| { |
| "name": "pi05_putcab_athenb_fullhorizon_mb16_ga1_lora3ep", |
| "project_name": "parallelvla-pi05-putcab-athenb-fullhorizon-autodl", |
| "exp_name": "<tyro._singleton.PropagatingMissingType object at 0x7f6675973790>", |
| "model": { |
| "action_dim": 32, |
| "action_horizon": 50, |
| "max_token_len": 200, |
| "dtype": "bfloat16", |
| "paligemma_variant": "gemma_2b_lora", |
| "action_expert_variant": "gemma_300m_lora", |
| "pi05": true, |
| "casm_mode": "none", |
| "coordination_gate_hidden_dim": 64, |
| "cross_attention_dim": 128, |
| "cross_output_rank": 4, |
| "skill_adapter_rank": 4, |
| "gate_loss_weight": 0.2, |
| "gate_positive_weight": 1.0, |
| "usefulness_loss_weight": 0.2, |
| "phase_prior_loss_weight": 0.1, |
| "usefulness_temperature": 0.1, |
| "semantic_subtask_prediction": false, |
| "semantic_role_loss_weight": 0.2, |
| "semantic_stage_loss_weight": 0.2, |
| "semantic_stage_class_weights": [ |
| 13.81, |
| 1.13, |
| 0.93, |
| 13.81, |
| 3.78, |
| 0.78 |
| ], |
| "discrete_state_input": true, |
| "bigate_sae": false, |
| "bigate_minimal_experts": false, |
| "bigate_joint_expert": false, |
| "bigate_inference_mode": "split", |
| "left_action_indices": [ |
| 0, |
| 1, |
| 2, |
| 3, |
| 4, |
| 5, |
| 6 |
| ], |
| "right_action_indices": [ |
| 7, |
| 8, |
| 9, |
| 10, |
| 11, |
| 12, |
| 13 |
| ], |
| "bigate_hidden_dim": 256, |
| "bigate_initial_probability": 0.99, |
| "bigate_gate_loss_weight": 0.1, |
| "bigate_communication_cost": 0.001, |
| "bigate_communication_dropout": 0.05, |
| "bigate_advantage_interval": 1, |
| "bigate_phase_a_steps": 1500, |
| "bigate_phase_b_end_step": 3500, |
| "bigate_training_gate_mode": "learned", |
| "bigate_inference_gate_mode": "learned" |
| }, |
| "weight_loader": { |
| "params_path": "/root/ctr-e025-performance/weights/pi05-base-source/params", |
| "missing_regex": ".*lora.*" |
| }, |
| "pytorch_weight_path": null, |
| "pytorch_training_precision": "bfloat16", |
| "pytorch_compute_precision": "float32", |
| "pytorch_attention_implementation": "sdpa", |
| "pytorch_gradient_checkpointing": true, |
| "pytorch_gradient_checkpointing_scope": "full", |
| "pytorch_fused_optimizer": false, |
| "pytorch_compile_mode": "none", |
| "pytorch_trainable_scope": "all", |
| "bigate_gate_freeze_steps": 0, |
| "bigate_joint_freeze_steps": 0, |
| "bigate_expert_lr": 1e-05, |
| "bigate_joint_expert_lr": 2e-06, |
| "bigate_projection_lr": 1e-05, |
| "bigate_gate_lr": 0.0001, |
| "lr_schedule": { |
| "warmup_steps": 1000, |
| "peak_lr": 2.5e-05, |
| "decay_steps": 30000, |
| "decay_lr": 2.5e-06 |
| }, |
| "optimizer": { |
| "b1": 0.9, |
| "b2": 0.95, |
| "eps": 1e-08, |
| "weight_decay": 1e-10, |
| "clip_gradient_norm": 1.0 |
| }, |
| "ema_decay": null, |
| "freeze_filter": "All(PathRegex(pattern=re.compile('.*llm.*'), sep='/'), Not(PathRegex(pattern=re.compile('.*lora.*'), sep='/')))", |
| "data": { |
| "repo_id": "Shiki42/PutCab-CTR-Train50-V4", |
| "assets": { |
| "assets_dir": "/root/autodl_oss_runs/putcab-reproduction-20260916/E229/norm-assets", |
| "asset_id": null |
| }, |
| "base_config": { |
| "repo_id": null, |
| "asset_id": null, |
| "norm_stats": null, |
| "repack_transforms": { |
| "inputs": [], |
| "outputs": [] |
| }, |
| "data_transforms": { |
| "inputs": [], |
| "outputs": [] |
| }, |
| "model_transforms": { |
| "inputs": [], |
| "outputs": [] |
| }, |
| "use_quantile_norm": false, |
| "action_sequence_keys": [ |
| "actions" |
| ], |
| "prompt_from_task": true, |
| "video_backend": "pyav", |
| "rlds_data_dir": null, |
| "action_space": null, |
| "filter_dict_path": null |
| }, |
| "use_delta_joint_actions": true, |
| "default_prompt": null, |
| "adapt_to_pi": false, |
| "inactive_action_weight": 0.0, |
| "repack_transforms": { |
| "inputs": [ |
| { |
| "structure": { |
| "images": { |
| "cam_high": "observation.images.cam_high", |
| "cam_left_wrist": "observation.images.cam_left_wrist", |
| "cam_right_wrist": "observation.images.cam_right_wrist" |
| }, |
| "state": "observation.state", |
| "actions": "action", |
| "action_is_pad": "action_is_pad", |
| "prompt": "prompt" |
| } |
| } |
| ], |
| "outputs": [] |
| }, |
| "action_sequence_keys": [ |
| "action" |
| ] |
| }, |
| "assets_base_dir": "./assets", |
| "checkpoint_base_dir": "./checkpoints", |
| "seed": 87431, |
| "train_episodes": null, |
| "validation_episodes": [], |
| "validation_interval": 0, |
| "validation_num_workers": 0, |
| "batch_size": 16, |
| "gradient_accumulation_steps": 1, |
| "num_workers": 4, |
| "prefetch_factor": 2, |
| "persistent_workers": true, |
| "pin_memory": true, |
| "num_train_steps": 10000, |
| "log_interval": 50, |
| "save_interval": 10000, |
| "keep_period": 10000, |
| "params_only_checkpoint": false, |
| "overwrite": false, |
| "resume": false, |
| "wandb_enabled": false, |
| "policy_metadata": null, |
| "fsdp_devices": 1 |
| } |
|
|
