Upload folder using huggingface_hub

outputs/2026-05-31/16-33-10/.hydra/config.yaml

@@ -0,0 +1,235 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 2000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 2000
+    val_every_n_epoch: null
+    limit_batch: 8
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+  n_frames_valid: 700
+  customized_validation: true
+  memory_condition_length: 0
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers: null
+  memory_condition_length: 0
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    debug_force_all_streams: false
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    token_patch_size: 2
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 3
+    dynamic:
+      enabled: true
+      exclude_latest_local_frames: 4
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      high_quality_fov_threshold: 0.7
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 3
+    stage_policy:
+      noise_bucket_logging: true
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      dit_hidden_size: 1024
+      anchor_gate: 1.0
+      dynamic_gate: 1.0
+      revisit_gate: 1.0
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: DeMemWM
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_full_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_full_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+vae_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-05-31/16-33-10/.hydra/hydra.yaml

@@ -0,0 +1,232 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_full_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_full_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+    - +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+    - dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=700
+    - +dataset.customized_validation=true
+    - +dataset.memory_condition_length=0
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.debug_force_all_streams=false
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=3
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.exclude_latest_local_frames=4
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=3
+    - ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=8
+    - experiment.training.checkpointing.every_n_train_steps=2000
+    - experiment.validation.val_every_n_step=2000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=3,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.debug_force_all_streams=false,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.exclude_latest_local_frames=4,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=3,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.dememwm.stage_policy.noise_bucket_logging=true,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.memory_condition_length=0,+dataset.n_frames_valid=700,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors,+name=train_dememwm_full_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_full_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features,dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=2000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=8,experiment.validation.val_every_n_step=2000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-05-31/16-33-10
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-05-31/16-33-10/.hydra/overrides.yaml

@@ -0,0 +1,74 @@
+- +name=train_dememwm_full_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_full_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+- +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+- dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=700
+- +dataset.customized_validation=true
+- +dataset.memory_condition_length=0
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.debug_force_all_streams=false
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=3
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.exclude_latest_local_frames=4
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=3
+- ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=8
+- experiment.training.checkpointing.every_n_train_steps=2000
+- experiment.validation.val_every_n_step=2000
+- experiment.training.max_steps=350000

outputs/2026-05-31/16-33-10/main.log

@@ -0,0 +1,38 @@
+[2026-05-31 16:34:05,141][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 7, MEMBER: 8/8
+[2026-05-31 16:34:13,901][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 4, MEMBER: 5/8
+[2026-05-31 16:34:18,601][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 5, MEMBER: 6/8
+[2026-05-31 16:34:18,897][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/8
+[2026-05-31 16:34:19,141][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/8
+[2026-05-31 16:34:19,288][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/8
+[2026-05-31 16:34:34,883][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 5 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-05-31 16:34:34,883][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 7 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-05-31 16:34:34,883][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 4 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-05-31 16:34:34,883][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 3 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-05-31 16:34:34,883][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 2 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-05-31 16:34:34,883][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 1 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-05-31 16:34:39,689][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-05-31 16:34:39,689][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-05-31 16:34:39,689][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-05-31 16:34:39,689][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-05-31 16:34:39,689][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-05-31 16:34:39,689][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-03 16:33:00,182][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 2] Received SIGTERM: 15
+[2026-06-03 16:33:00,183][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 7] Received SIGTERM: 15
+[2026-06-03 16:33:00,184][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:00,182][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 1] Received SIGTERM: 15
+[2026-06-03 16:33:00,184][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:00,184][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:00,183][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 4] Received SIGTERM: 15
+[2026-06-03 16:33:00,186][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:00,189][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 1] Received SIGTERM: 15
+[2026-06-03 16:33:00,189][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:00,189][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 2] Received SIGTERM: 15
+[2026-06-03 16:33:00,189][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:00,189][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 4] Received SIGTERM: 15
+[2026-06-03 16:33:00,189][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:00,189][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 7] Received SIGTERM: 15
+[2026-06-03 16:33:00,190][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:01,747][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 5] Received SIGTERM: 15
+[2026-06-03 16:33:01,748][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:01,748][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 3] Received SIGTERM: 15
+[2026-06-03 16:33:01,749][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15

outputs/2026-05-31/16-33-11/.hydra/config.yaml

@@ -0,0 +1,235 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 2000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 2000
+    val_every_n_epoch: null
+    limit_batch: 8
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+  n_frames_valid: 700
+  customized_validation: true
+  memory_condition_length: 0
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers: null
+  memory_condition_length: 0
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    debug_force_all_streams: false
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    token_patch_size: 2
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 3
+    dynamic:
+      enabled: true
+      exclude_latest_local_frames: 4
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      high_quality_fov_threshold: 0.7
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 3
+    stage_policy:
+      noise_bucket_logging: true
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      dit_hidden_size: 1024
+      anchor_gate: 1.0
+      dynamic_gate: 1.0
+      revisit_gate: 1.0
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: DeMemWM
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_full_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_full_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+vae_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-05-31/16-33-11/.hydra/hydra.yaml

@@ -0,0 +1,232 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_full_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_full_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+    - +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+    - dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=700
+    - +dataset.customized_validation=true
+    - +dataset.memory_condition_length=0
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.debug_force_all_streams=false
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=3
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.exclude_latest_local_frames=4
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=3
+    - ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=8
+    - experiment.training.checkpointing.every_n_train_steps=2000
+    - experiment.validation.val_every_n_step=2000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=3,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.debug_force_all_streams=false,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.exclude_latest_local_frames=4,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=3,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.dememwm.stage_policy.noise_bucket_logging=true,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.memory_condition_length=0,+dataset.n_frames_valid=700,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors,+name=train_dememwm_full_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_full_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features,dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=2000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=8,experiment.validation.val_every_n_step=2000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-05-31/16-33-11
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-05-31/16-33-11/.hydra/overrides.yaml

@@ -0,0 +1,74 @@
+- +name=train_dememwm_full_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_full_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+- +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+- dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=700
+- +dataset.customized_validation=true
+- +dataset.memory_condition_length=0
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.debug_force_all_streams=false
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=3
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.exclude_latest_local_frames=4
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=3
+- ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=8
+- experiment.training.checkpointing.every_n_train_steps=2000
+- experiment.validation.val_every_n_step=2000
+- experiment.training.max_steps=350000

outputs/2026-05-31/16-33-11/main.log

@@ -0,0 +1,39 @@
+[2026-05-31 16:33:58,507][pytorch_lightning.utilities.rank_zero][INFO] - Using 16bit Automatic Mixed Precision (AMP)
+[2026-05-31 16:33:58,787][pytorch_lightning.utilities.rank_zero][INFO] - GPU available: True (cuda), used: True
+[2026-05-31 16:33:58,787][pytorch_lightning.utilities.rank_zero][INFO] - TPU available: False, using: 0 TPU cores
+[2026-05-31 16:33:58,787][pytorch_lightning.utilities.rank_zero][INFO] - IPU available: False, using: 0 IPUs
+[2026-05-31 16:33:58,787][pytorch_lightning.utilities.rank_zero][INFO] - HPU available: False, using: 0 HPUs
+[2026-05-31 16:34:01,148][pytorch_lightning.utilities.rank_zero][INFO] - Model weights loaded from /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors: matched=302 missing=0 shape_mismatch=0 prefix=''
+[2026-05-31 16:34:02,180][pytorch_lightning.utilities.rank_zero][INFO] - Model weights loaded from /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors: matched=228 missing=0 shape_mismatch=0 prefix=''
+[2026-05-31 16:34:05,142][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/8
+[2026-05-31 16:34:14,613][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 6, MEMBER: 7/8
+[2026-05-31 16:34:20,380][pytorch_lightning.utilities.rank_zero][INFO] - ----------------------------------------------------------------------------------------------------
+distributed_backend=nccl
+All distributed processes registered. Starting with 8 processes
+----------------------------------------------------------------------------------------------------
+
+[2026-05-31 16:34:34,882][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-05-31 16:34:34,883][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 6 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-05-31 16:34:36,691][lightning.pytorch.callbacks.model_summary][INFO] - 
+  | Name                       | Type                                  | Params
+-------------------------------------------------------------------------------------
+0 | diffusion_model            | Diffusion                             | 910 M 
+1 | dememwm_dynamic_compressor | CausalConv3DDynamicCompressor         | 199 K 
+2 | dememwm_anchor_proj        | SpatialConv2DMemoryProjector          | 591 K 
+3 | dememwm_revisit_proj       | SpatialConv2DMemoryProjector          | 591 K 
+4 | dememwm_revisit_gate       | RevisitRawGate                        | 4     
+5 | validation_lpips_model     | LearnedPerceptualImagePatchSimilarity | 2.5 M 
+6 | vae                        | AutoencoderKL                         | 229 M 
+-------------------------------------------------------------------------------------
+911 M     Trainable params
+231 M     Non-trainable params
+1.1 B     Total params
+4,573.841 Total estimated model params size (MB)
+[2026-05-31 16:34:39,689][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-05-31 16:34:39,689][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-03 16:33:00,183][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 6] Received SIGTERM: 15
+[2026-06-03 16:33:00,184][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:00,189][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 6] Received SIGTERM: 15
+[2026-06-03 16:33:00,189][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-03 16:33:01,749][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 0] Received SIGTERM: 15
+[2026-06-03 16:33:01,750][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15

outputs/2026-06-01/15-41-01/.hydra/config.yaml

@@ -0,0 +1,234 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 2000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 2000
+    val_every_n_epoch: null
+    limit_batch: 2
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /share_1/users/bonan_ding/worldmem_data/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+  n_frames_valid: 1100
+  customized_validation: true
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers:
+  - 3
+  - 7
+  - 11
+  - 15
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    debug_force_all_streams: false
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    spatial_kernel_size: 3
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 2
+    dynamic:
+      enabled: true
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+      compress:
+        downsample_ratio: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 2
+    stage_policy:
+      noise_bucket_logging: true
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      memory_hidden_size: 512
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+vae_path: /share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-01/15-41-01/.hydra/hydra.yaml

@@ -0,0 +1,232 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+    - +vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft
+    - dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=1100
+    - +dataset.customized_validation=true
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.debug_force_all_streams=false
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.dynamic.conv_stride_t=2
+    - ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+    - ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=2
+    - experiment.training.checkpointing.every_n_train_steps=2000
+    - experiment.validation.val_every_n_step=2000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=2,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.debug_force_all_streams=false,++algorithm.dememwm.dynamic.compress.downsample_ratio=2,++algorithm.dememwm.dynamic.conv_stride_t=2,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=2,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.dememwm.stage_policy.noise_bucket_logging=true,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.n_frames_valid=1100,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors,+name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features,dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=2000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=2,experiment.validation.val_every_n_step=2000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-01/15-41-01
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-06-01/15-41-01/.hydra/overrides.yaml

@@ -0,0 +1,74 @@
+- +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+- +vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft
+- dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=1100
+- +dataset.customized_validation=true
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.debug_force_all_streams=false
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.dynamic.conv_stride_t=2
+- ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+- ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=2
+- experiment.training.checkpointing.every_n_train_steps=2000
+- experiment.validation.val_every_n_step=2000
+- experiment.training.max_steps=350000

outputs/2026-06-01/15-41-06/.hydra/config.yaml

@@ -0,0 +1,234 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 2000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 2000
+    val_every_n_epoch: null
+    limit_batch: 2
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /share_1/users/bonan_ding/worldmem_data/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+  n_frames_valid: 1100
+  customized_validation: true
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers:
+  - 3
+  - 7
+  - 11
+  - 15
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    debug_force_all_streams: false
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    spatial_kernel_size: 3
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 2
+    dynamic:
+      enabled: true
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+      compress:
+        downsample_ratio: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 2
+    stage_policy:
+      noise_bucket_logging: true
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      memory_hidden_size: 512
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+vae_path: /share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-01/15-41-06/.hydra/hydra.yaml

@@ -0,0 +1,232 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+    - +vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft
+    - dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=1100
+    - +dataset.customized_validation=true
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.debug_force_all_streams=false
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.dynamic.conv_stride_t=2
+    - ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+    - ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=2
+    - experiment.training.checkpointing.every_n_train_steps=2000
+    - experiment.validation.val_every_n_step=2000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=2,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.debug_force_all_streams=false,++algorithm.dememwm.dynamic.compress.downsample_ratio=2,++algorithm.dememwm.dynamic.conv_stride_t=2,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=2,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.dememwm.stage_policy.noise_bucket_logging=true,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.n_frames_valid=1100,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors,+name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features,dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=2000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=2,experiment.validation.val_every_n_step=2000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-01/15-41-06
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-06-01/15-41-06/.hydra/overrides.yaml

@@ -0,0 +1,74 @@
+- +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+- +vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft
+- dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=1100
+- +dataset.customized_validation=true
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.debug_force_all_streams=false
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.dynamic.conv_stride_t=2
+- ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+- ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=2
+- experiment.training.checkpointing.every_n_train_steps=2000
+- experiment.validation.val_every_n_step=2000
+- experiment.training.max_steps=350000

outputs/2026-06-01/15-41-07/.hydra/config.yaml

@@ -0,0 +1,234 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 2000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 2000
+    val_every_n_epoch: null
+    limit_batch: 2
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /share_1/users/bonan_ding/worldmem_data/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+  n_frames_valid: 1100
+  customized_validation: true
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers:
+  - 3
+  - 7
+  - 11
+  - 15
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    debug_force_all_streams: false
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    spatial_kernel_size: 3
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 2
+    dynamic:
+      enabled: true
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+      compress:
+        downsample_ratio: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 2
+    stage_policy:
+      noise_bucket_logging: true
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      memory_hidden_size: 512
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+vae_path: /share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-01/15-41-07/.hydra/hydra.yaml

@@ -0,0 +1,232 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+    - +vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft
+    - dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=1100
+    - +dataset.customized_validation=true
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.debug_force_all_streams=false
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.dynamic.conv_stride_t=2
+    - ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+    - ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=2
+    - experiment.training.checkpointing.every_n_train_steps=2000
+    - experiment.validation.val_every_n_step=2000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=2,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.debug_force_all_streams=false,++algorithm.dememwm.dynamic.compress.downsample_ratio=2,++algorithm.dememwm.dynamic.conv_stride_t=2,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=2,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.dememwm.stage_policy.noise_bucket_logging=true,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.n_frames_valid=1100,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors,+name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features,dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=2000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=2,experiment.validation.val_every_n_step=2000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-01/15-41-07
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-06-01/15-41-07/.hydra/overrides.yaml

@@ -0,0 +1,74 @@
+- +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+- +vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft
+- dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=1100
+- +dataset.customized_validation=true
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.debug_force_all_streams=false
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.dynamic.conv_stride_t=2
+- ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+- ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=2
+- experiment.training.checkpointing.every_n_train_steps=2000
+- experiment.validation.val_every_n_step=2000
+- experiment.training.max_steps=350000

outputs/2026-06-01/15-41-08/.hydra/config.yaml

@@ -0,0 +1,234 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 2000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 2000
+    val_every_n_epoch: null
+    limit_batch: 2
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /share_1/users/bonan_ding/worldmem_data/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+  n_frames_valid: 1100
+  customized_validation: true
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers:
+  - 3
+  - 7
+  - 11
+  - 15
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    debug_force_all_streams: false
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    spatial_kernel_size: 3
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 2
+    dynamic:
+      enabled: true
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+      compress:
+        downsample_ratio: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 2
+    stage_policy:
+      noise_bucket_logging: true
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      memory_hidden_size: 512
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+vae_path: /share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-01/15-41-08/.hydra/hydra.yaml

@@ -0,0 +1,232 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+    - +vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft
+    - dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=1100
+    - +dataset.customized_validation=true
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.debug_force_all_streams=false
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.dynamic.conv_stride_t=2
+    - ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+    - ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=2
+    - experiment.training.checkpointing.every_n_train_steps=2000
+    - experiment.validation.val_every_n_step=2000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=2,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.debug_force_all_streams=false,++algorithm.dememwm.dynamic.compress.downsample_ratio=2,++algorithm.dememwm.dynamic.conv_stride_t=2,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=2,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.dememwm.stage_policy.noise_bucket_logging=true,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.n_frames_valid=1100,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors,+name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features,dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=2000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=2,experiment.validation.val_every_n_step=2000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-01/15-41-08
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-06-01/15-41-08/.hydra/overrides.yaml

@@ -0,0 +1,74 @@
+- +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/oasis500m.safetensors
+- +vae_path=/share_1/users/bonan_ding/WorldMem/open-oasis/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/share_1/users/bonan_ding/worldmem_data/minecraft
+- dataset.precomputed_feature_dir=/share_1/users/bonan_ding/worldmem_data/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=1100
+- +dataset.customized_validation=true
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.debug_force_all_streams=false
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.dynamic.conv_stride_t=2
+- ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+- ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=2
+- experiment.training.checkpointing.every_n_train_steps=2000
+- experiment.validation.val_every_n_step=2000
+- experiment.training.max_steps=350000

outputs/2026-06-01/15-41-08/main.log

@@ -0,0 +1,5 @@
+[2026-06-01 15:41:49,208][pytorch_lightning.utilities.rank_zero][INFO] - Using 16bit Automatic Mixed Precision (AMP)
+[2026-06-01 15:41:49,334][pytorch_lightning.utilities.rank_zero][INFO] - GPU available: True (cuda), used: True
+[2026-06-01 15:41:49,334][pytorch_lightning.utilities.rank_zero][INFO] - TPU available: False, using: 0 TPU cores
+[2026-06-01 15:41:49,334][pytorch_lightning.utilities.rank_zero][INFO] - IPU available: False, using: 0 IPUs
+[2026-06-01 15:41:49,334][pytorch_lightning.utilities.rank_zero][INFO] - HPU available: False, using: 0 HPUs

outputs/2026-06-03/20-22-54/.hydra/config.yaml

@@ -0,0 +1,234 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 2000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 2000
+    val_every_n_epoch: null
+    limit_batch: 2
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+  n_frames_valid: 1100
+  customized_validation: true
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers:
+  - 3
+  - 7
+  - 11
+  - 15
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    debug_force_all_streams: false
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    spatial_kernel_size: 3
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 2
+    dynamic:
+      enabled: true
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+      compress:
+        downsample_ratio: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 2
+    stage_policy:
+      noise_bucket_logging: true
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      memory_hidden_size: 512
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+vae_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-03/20-22-54/.hydra/hydra.yaml

@@ -0,0 +1,232 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+    - +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+    - dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=1100
+    - +dataset.customized_validation=true
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.debug_force_all_streams=false
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.dynamic.conv_stride_t=2
+    - ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+    - ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=2
+    - experiment.training.checkpointing.every_n_train_steps=2000
+    - experiment.validation.val_every_n_step=2000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=2,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.debug_force_all_streams=false,++algorithm.dememwm.dynamic.compress.downsample_ratio=2,++algorithm.dememwm.dynamic.conv_stride_t=2,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=2,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.dememwm.stage_policy.noise_bucket_logging=true,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.n_frames_valid=1100,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors,+name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features,dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=2000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=2,experiment.validation.val_every_n_step=2000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-03/20-22-54
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-06-03/20-22-54/.hydra/overrides.yaml

@@ -0,0 +1,74 @@
+- +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+- +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+- dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=1100
+- +dataset.customized_validation=true
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.debug_force_all_streams=false
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.dynamic.conv_stride_t=2
+- ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+- ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=2
+- experiment.training.checkpointing.every_n_train_steps=2000
+- experiment.validation.val_every_n_step=2000
+- experiment.training.max_steps=350000

outputs/2026-06-03/20-22-54/main.log

@@ -0,0 +1,47 @@
+[2026-06-03 20:23:36,031][pytorch_lightning.utilities.rank_zero][INFO] - Using 16bit Automatic Mixed Precision (AMP)
+[2026-06-03 20:23:36,800][pytorch_lightning.utilities.rank_zero][INFO] - GPU available: True (cuda), used: True
+[2026-06-03 20:23:36,800][pytorch_lightning.utilities.rank_zero][INFO] - TPU available: False, using: 0 TPU cores
+[2026-06-03 20:23:36,800][pytorch_lightning.utilities.rank_zero][INFO] - IPU available: False, using: 0 IPUs
+[2026-06-03 20:23:36,800][pytorch_lightning.utilities.rank_zero][INFO] - HPU available: False, using: 0 HPUs
+[2026-06-03 20:23:38,339][pytorch_lightning.utilities.rank_zero][INFO] - Model weights loaded from /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors: matched=302 missing=0 shape_mismatch=0 prefix=''
+[2026-06-03 20:23:39,236][pytorch_lightning.utilities.rank_zero][INFO] - Model weights loaded from /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors: matched=228 missing=0 shape_mismatch=0 prefix=''
+[2026-06-03 20:23:42,001][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/8
+[2026-06-03 20:23:42,005][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/8
+[2026-06-03 20:23:42,015][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/8
+[2026-06-03 20:23:42,026][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/8
+[2026-06-03 20:23:48,581][pytorch_lightning.utilities.rank_zero][INFO] - ----------------------------------------------------------------------------------------------------
+distributed_backend=nccl
+All distributed processes registered. Starting with 8 processes
+----------------------------------------------------------------------------------------------------
+
+[2026-06-03 20:24:03,625][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-03 20:24:03,625][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 1 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-03 20:24:03,625][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 2 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-03 20:24:03,625][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 3 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-03 20:24:05,745][lightning.pytorch.callbacks.model_summary][INFO] - 
+  | Name                       | Type                                  | Params
+-------------------------------------------------------------------------------------
+0 | diffusion_model            | Diffusion                             | 631 M 
+1 | dememwm_dynamic_compressor | CausalConv3DDynamicCompressor         | 887 K 
+2 | dememwm_anchor_proj        | SpatialConv2DMemoryProjector          | 296 K 
+3 | dememwm_revisit_proj       | SpatialConv2DMemoryProjector          | 296 K 
+4 | dememwm_revisit_gate       | RevisitRawGate                        | 4     
+5 | validation_lpips_model     | LearnedPerceptualImagePatchSimilarity | 2.5 M 
+6 | vae                        | AutoencoderKL                         | 229 M 
+-------------------------------------------------------------------------------------
+632 M     Trainable params
+231 M     Non-trainable params
+864 M     Total params
+3,456.995 Total estimated model params size (MB)
+[2026-06-03 20:24:08,742][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-03 20:24:08,742][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-03 20:24:08,742][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-03 20:24:08,742][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-04 06:21:13,812][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 3] Received SIGTERM: 15
+[2026-06-04 06:21:13,813][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-04 06:21:13,812][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 2] Received SIGTERM: 15
+[2026-06-04 06:21:13,813][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-04 06:21:13,812][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 1] Received SIGTERM: 15
+[2026-06-04 06:21:13,812][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 0] Received SIGTERM: 15
+[2026-06-04 06:21:13,814][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-04 06:21:13,814][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15

outputs/2026-06-03/20-22-55/.hydra/config.yaml

@@ -0,0 +1,234 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 2000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 2000
+    val_every_n_epoch: null
+    limit_batch: 2
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+  n_frames_valid: 1100
+  customized_validation: true
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers:
+  - 3
+  - 7
+  - 11
+  - 15
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    debug_force_all_streams: false
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    spatial_kernel_size: 3
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 2
+    dynamic:
+      enabled: true
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+      compress:
+        downsample_ratio: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 2
+    stage_policy:
+      noise_bucket_logging: true
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      memory_hidden_size: 512
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+vae_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-03/20-22-55/.hydra/hydra.yaml

@@ -0,0 +1,232 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+    - +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+    - dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=1100
+    - +dataset.customized_validation=true
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.debug_force_all_streams=false
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.dynamic.conv_stride_t=2
+    - ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+    - ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=2
+    - experiment.training.checkpointing.every_n_train_steps=2000
+    - experiment.validation.val_every_n_step=2000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=2,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.debug_force_all_streams=false,++algorithm.dememwm.dynamic.compress.downsample_ratio=2,++algorithm.dememwm.dynamic.conv_stride_t=2,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=2,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.dememwm.stage_policy.noise_bucket_logging=true,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.n_frames_valid=1100,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors,+name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features,dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=2000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=2,experiment.validation.val_every_n_step=2000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-03/20-22-55
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-06-03/20-22-55/.hydra/overrides.yaml

@@ -0,0 +1,74 @@
+- +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+- +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+- dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=1100
+- +dataset.customized_validation=true
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.debug_force_all_streams=false
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.dynamic.conv_stride_t=2
+- ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+- ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=2
+- experiment.training.checkpointing.every_n_train_steps=2000
+- experiment.validation.val_every_n_step=2000
+- experiment.training.max_steps=350000

outputs/2026-06-03/20-22-55/main.log

@@ -0,0 +1,10 @@
+[2026-06-03 20:23:41,996][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 6, MEMBER: 7/8
+[2026-06-03 20:23:42,011][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 4, MEMBER: 5/8
+[2026-06-03 20:24:03,625][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 4 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-03 20:24:03,625][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 6 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-03 20:24:08,742][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-03 20:24:08,742][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-04 06:21:13,812][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 4] Received SIGTERM: 15
+[2026-06-04 06:21:13,813][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-04 06:21:13,812][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 6] Received SIGTERM: 15
+[2026-06-04 06:21:13,813][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15

outputs/2026-06-03/20-22-56/.hydra/config.yaml

@@ -0,0 +1,234 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 2000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 2000
+    val_every_n_epoch: null
+    limit_batch: 2
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+  n_frames_valid: 1100
+  customized_validation: true
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers:
+  - 3
+  - 7
+  - 11
+  - 15
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    debug_force_all_streams: false
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    spatial_kernel_size: 3
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 2
+    dynamic:
+      enabled: true
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+      compress:
+        downsample_ratio: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 2
+    stage_policy:
+      noise_bucket_logging: true
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      memory_hidden_size: 512
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+vae_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-03/20-22-56/.hydra/hydra.yaml

@@ -0,0 +1,232 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+    - +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+    - dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=1100
+    - +dataset.customized_validation=true
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.debug_force_all_streams=false
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.dynamic.conv_stride_t=2
+    - ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+    - ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=2
+    - experiment.training.checkpointing.every_n_train_steps=2000
+    - experiment.validation.val_every_n_step=2000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=2,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.debug_force_all_streams=false,++algorithm.dememwm.dynamic.compress.downsample_ratio=2,++algorithm.dememwm.dynamic.conv_stride_t=2,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=2,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.dememwm.stage_policy.noise_bucket_logging=true,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.n_frames_valid=1100,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors,+name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features,dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=2000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=2,experiment.validation.val_every_n_step=2000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-03/20-22-56
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-06-03/20-22-56/.hydra/overrides.yaml

@@ -0,0 +1,74 @@
+- +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+- +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+- dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=1100
+- +dataset.customized_validation=true
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.debug_force_all_streams=false
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.dynamic.conv_stride_t=2
+- ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+- ++algorithm.dememwm.stage_policy.noise_bucket_logging=true
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=2
+- experiment.training.checkpointing.every_n_train_steps=2000
+- experiment.validation.val_every_n_step=2000
+- experiment.training.max_steps=350000

outputs/2026-06-03/20-22-56/main.log

@@ -0,0 +1,10 @@
+[2026-06-03 20:23:42,010][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 7, MEMBER: 8/8
+[2026-06-03 20:23:42,017][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 5, MEMBER: 6/8
+[2026-06-03 20:24:03,625][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 7 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-03 20:24:03,625][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 5 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-03 20:24:08,742][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-03 20:24:08,742][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-04 06:21:13,812][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 5] Received SIGTERM: 15
+[2026-06-04 06:21:13,813][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-04 06:21:13,813][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 7] Received SIGTERM: 15
+[2026-06-04 06:21:13,813][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15

outputs/2026-06-04/06-22-27/.hydra/config.yaml

@@ -0,0 +1,231 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 5000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 5000
+    val_every_n_epoch: null
+    limit_batch: 1
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+  n_frames_valid: 500
+  customized_validation: true
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers:
+  - 3
+  - 7
+  - 11
+  - 15
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    spatial_kernel_size: 3
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 2
+    dynamic:
+      enabled: true
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+      compress:
+        downsample_ratio: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 2
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      memory_hidden_size: 512
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+vae_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-04/06-22-27/.hydra/hydra.yaml

@@ -0,0 +1,230 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+    - +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+    - dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=500
+    - +dataset.customized_validation=true
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.dynamic.conv_stride_t=2
+    - ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=1
+    - experiment.training.checkpointing.every_n_train_steps=5000
+    - experiment.validation.val_every_n_step=5000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=2,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.dynamic.compress.downsample_ratio=2,++algorithm.dememwm.dynamic.conv_stride_t=2,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=2,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.n_frames_valid=500,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors,+name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features,dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=5000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=1,experiment.validation.val_every_n_step=5000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-04/06-22-27
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-06-04/06-22-27/.hydra/overrides.yaml

@@ -0,0 +1,72 @@
+- +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+- +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+- dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=500
+- +dataset.customized_validation=true
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.dynamic.conv_stride_t=2
+- ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=1
+- experiment.training.checkpointing.every_n_train_steps=5000
+- experiment.validation.val_every_n_step=5000
+- experiment.training.max_steps=350000

outputs/2026-06-04/06-22-28/.hydra/config.yaml

@@ -0,0 +1,231 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 5000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 5000
+    val_every_n_epoch: null
+    limit_batch: 1
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+  n_frames_valid: 500
+  customized_validation: true
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers:
+  - 3
+  - 7
+  - 11
+  - 15
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    spatial_kernel_size: 3
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 2
+    dynamic:
+      enabled: true
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+      compress:
+        downsample_ratio: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 2
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      memory_hidden_size: 512
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+vae_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-04/06-22-28/.hydra/hydra.yaml

@@ -0,0 +1,230 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+    - +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+    - dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=500
+    - +dataset.customized_validation=true
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.dynamic.conv_stride_t=2
+    - ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=1
+    - experiment.training.checkpointing.every_n_train_steps=5000
+    - experiment.validation.val_every_n_step=5000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=2,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.dynamic.compress.downsample_ratio=2,++algorithm.dememwm.dynamic.conv_stride_t=2,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=2,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.n_frames_valid=500,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors,+name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features,dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=5000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=1,experiment.validation.val_every_n_step=5000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-04/06-22-28
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-06-04/06-22-28/.hydra/overrides.yaml

@@ -0,0 +1,72 @@
+- +name=train_dememwm_separate_memory_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_separate_memory_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+- +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+- dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=500
+- +dataset.customized_validation=true
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=2
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.dynamic.conv_stride_t=2
+- ++algorithm.dememwm.dynamic.compress.downsample_ratio=2
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=2
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=1
+- experiment.training.checkpointing.every_n_train_steps=5000
+- experiment.validation.val_every_n_step=5000
+- experiment.training.max_steps=350000

outputs/2026-06-04/06-24-27/.hydra/config.yaml

@@ -0,0 +1,232 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 5000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 5000
+    val_every_n_epoch: null
+    limit_batch: 1
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+  n_frames_valid: 500
+  customized_validation: true
+  memory_condition_length: 0
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers: null
+  memory_condition_length: 0
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    token_patch_size: 2
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 3
+    dynamic:
+      enabled: true
+      exclude_latest_local_frames: 4
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      high_quality_fov_threshold: 0.7
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 3
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      dit_hidden_size: 1024
+      anchor_gate: 1.0
+      dynamic_gate: 1.0
+      revisit_gate: 1.0
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_full_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_full_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+vae_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-04/06-24-27/.hydra/hydra.yaml

@@ -0,0 +1,230 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_full_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_full_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+    - +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+    - dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=500
+    - +dataset.customized_validation=true
+    - +dataset.memory_condition_length=0
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=3
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.exclude_latest_local_frames=4
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=3
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=1
+    - experiment.training.checkpointing.every_n_train_steps=5000
+    - experiment.validation.val_every_n_step=5000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=3,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.exclude_latest_local_frames=4,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=3,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.memory_condition_length=0,+dataset.n_frames_valid=500,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors,+name=train_dememwm_full_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_full_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features,dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=5000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=1,experiment.validation.val_every_n_step=5000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-04/06-24-27
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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

outputs/2026-06-04/06-24-27/.hydra/overrides.yaml

@@ -0,0 +1,72 @@
+- +name=train_dememwm_full_berzelius_8a100_bs16_global128_350k
+- +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_full_berzelius_8a100_bs16_global128_350k/
+- auto_resume=true
+- experiment.tasks=[training]
+- algorithm=dememwm_memory_dit
+- +customized_load=true
+- +seperate_load=true
+- +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+- +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+- +only_tune_memory=false
+- dataset=video_minecraft_latent
+- dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+- dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+- dataset.n_frames=1000
+- +dataset.n_frames_valid=500
+- +dataset.customized_validation=true
+- +dataset.memory_condition_length=0
+- +dataset.wo_updown=false
+- +dataset.angle_range=180
+- +dataset.pos_range=8
+- ++algorithm.n_tokens=4
+- algorithm.x_shape=[16,18,32]
+- ++algorithm.context_frames=100
+- ++algorithm.log_video=true
+- ++algorithm.diffusion.sampling_timesteps=20
+- ++algorithm.dememwm.generated_history_proxy.enabled=true
+- ++algorithm.dememwm.generated_history_proxy.start_step=40000
+- ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+- ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+- ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+- ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+- ++algorithm.dememwm.anchor.enabled=true
+- ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+- ++algorithm.dememwm.anchor.diverse_selection=true
+- ++algorithm.dememwm.anchor.compress.downsample_ratio=3
+- ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+- ++algorithm.dememwm.dynamic.enabled=true
+- ++algorithm.dememwm.dynamic.exclude_latest_local_frames=4
+- ++algorithm.dememwm.dynamic.recent_frames=4
+- ++algorithm.dememwm.revisit.enabled=true
+- ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+- ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+- ++algorithm.dememwm.revisit.pose_preselect_topk=64
+- ++algorithm.dememwm.revisit.fov_yaw_samples=25
+- ++algorithm.dememwm.revisit.fov_pitch_samples=20
+- ++algorithm.dememwm.revisit.fov_depth_samples=20
+- ++algorithm.dememwm.revisit.plucker_weight=0.10
+- ++algorithm.dememwm.revisit.max_frames=2
+- ++algorithm.dememwm.revisit.compress.downsample_ratio=3
+- ++algorithm.dememwm.cache.enabled=true
+- ++algorithm.dememwm.cache.device=cpu
+- ++algorithm.dememwm.cache.keep_raw_latents=all
+- ++algorithm.dememwm.cache.keep_compressed_records=true
+- ++algorithm.dememwm.cache.eviction_policy=none
+- ++algorithm.dememwm.cache.no_evict=true
+- ++algorithm.dememwm.cache.clear_between_videos=true
+- ++algorithm.dememwm.cache.max_records=null
+- ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+- ++algorithm.dememwm.curriculum.enabled=true
+- ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+- ++algorithm.dememwm.curriculum.freeze_vae=true
+- ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+- ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+- ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+- experiment.training.batch_size=16
+- experiment.training.optim.accumulate_grad_batches=1
+- experiment.validation.batch_size=1
+- experiment.validation.limit_batch=1
+- experiment.training.checkpointing.every_n_train_steps=5000
+- experiment.validation.val_every_n_step=5000
+- experiment.training.max_steps=350000

outputs/2026-06-04/06-24-27/main.log

@@ -0,0 +1,77 @@
+[2026-06-04 06:24:58,261][pytorch_lightning.utilities.rank_zero][INFO] - Using 16bit Automatic Mixed Precision (AMP)
+[2026-06-04 06:24:58,374][pytorch_lightning.utilities.rank_zero][INFO] - GPU available: True (cuda), used: True
+[2026-06-04 06:24:58,374][pytorch_lightning.utilities.rank_zero][INFO] - TPU available: False, using: 0 TPU cores
+[2026-06-04 06:24:58,374][pytorch_lightning.utilities.rank_zero][INFO] - IPU available: False, using: 0 IPUs
+[2026-06-04 06:24:58,374][pytorch_lightning.utilities.rank_zero][INFO] - HPU available: False, using: 0 HPUs
+[2026-06-04 06:24:58,374][pytorch_lightning.utilities.rank_zero][INFO] - `Trainer(limit_val_batches=1)` was configured so 1 batch will be used.
+[2026-06-04 06:24:58,617][pytorch_lightning.utilities.rank_zero][INFO] - Model weights loaded from /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors: matched=302 missing=0 shape_mismatch=0 prefix=''
+[2026-06-04 06:24:58,844][pytorch_lightning.utilities.rank_zero][INFO] - Model weights loaded from /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors: matched=228 missing=0 shape_mismatch=0 prefix=''
+[2026-06-04 06:25:01,130][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/8
+[2026-06-04 06:25:01,209][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 7, MEMBER: 8/8
+[2026-06-04 06:25:01,432][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/8
+[2026-06-04 06:25:01,930][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 6, MEMBER: 7/8
+[2026-06-04 06:25:02,129][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 4, MEMBER: 5/8
+[2026-06-04 06:25:02,209][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/8
+[2026-06-04 06:25:03,309][lightning.fabric.utilities.distributed][INFO] - Initializing distributed: GLOBAL_RANK: 5, MEMBER: 6/8
+[2026-06-04 06:25:07,333][pytorch_lightning.utilities.rank_zero][INFO] - ----------------------------------------------------------------------------------------------------
+distributed_backend=nccl
+All distributed processes registered. Starting with 8 processes
+----------------------------------------------------------------------------------------------------
+
+[2026-06-04 06:25:20,836][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-04 06:25:20,836][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 2 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-04 06:25:20,836][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 4 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-04 06:25:20,836][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 6 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-04 06:25:20,836][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 7 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-04 06:25:20,836][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 3 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-04 06:25:20,837][lightning.pytorch.accelerators.cuda][INFO] - LOCAL_RANK: 5 - CUDA_VISIBLE_DEVICES: [0,1,2,3,4,5,6,7]
+[2026-06-04 06:25:21,907][lightning.pytorch.callbacks.model_summary][INFO] - 
+  | Name                       | Type                                  | Params
+-------------------------------------------------------------------------------------
+0 | diffusion_model            | Diffusion                             | 910 M 
+1 | dememwm_dynamic_compressor | CausalConv3DDynamicCompressor         | 199 K 
+2 | dememwm_anchor_proj        | SpatialConv2DMemoryProjector          | 591 K 
+3 | dememwm_revisit_proj       | SpatialConv2DMemoryProjector          | 591 K 
+4 | dememwm_revisit_gate       | RevisitRawGate                        | 4     
+5 | validation_lpips_model     | LearnedPerceptualImagePatchSimilarity | 2.5 M 
+6 | vae                        | AutoencoderKL                         | 229 M 
+-------------------------------------------------------------------------------------
+911 M     Trainable params
+231 M     Non-trainable params
+1.1 B     Total params
+4,573.841 Total estimated model params size (MB)
+[2026-06-04 06:25:22,815][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-04 06:25:22,815][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-04 06:25:22,815][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-04 06:25:22,815][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-04 06:25:22,815][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-04 06:25:22,815][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-04 06:25:22,815][lightning.pytorch.trainer.connectors.signal_connector][INFO] - SLURM auto-requeueing enabled. Setting signal handlers.
+[2026-06-07 06:24:48,015][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 2] Received SIGTERM: 15
+[2026-06-07 06:24:48,015][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 3] Received SIGTERM: 15
+[2026-06-07 06:24:48,016][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,015][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 4] Received SIGTERM: 15
+[2026-06-07 06:24:48,016][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,015][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 5] Received SIGTERM: 15
+[2026-06-07 06:24:48,016][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,016][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,015][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 6] Received SIGTERM: 15
+[2026-06-07 06:24:48,016][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,016][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 7] Received SIGTERM: 15
+[2026-06-07 06:24:48,015][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 0] Received SIGTERM: 15
+[2026-06-07 06:24:48,016][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,016][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,020][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 2] Received SIGTERM: 15
+[2026-06-07 06:24:48,020][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 0] Received SIGTERM: 15
+[2026-06-07 06:24:48,020][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,020][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 3] Received SIGTERM: 15
+[2026-06-07 06:24:48,020][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,020][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 4] Received SIGTERM: 15
+[2026-06-07 06:24:48,020][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,020][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,020][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 5] Received SIGTERM: 15
+[2026-06-07 06:24:48,021][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,021][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 6] Received SIGTERM: 15
+[2026-06-07 06:24:48,021][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15
+[2026-06-07 06:24:48,021][lightning.pytorch.trainer.connectors.signal_connector][INFO] - [rank: 7] Received SIGTERM: 15
+[2026-06-07 06:24:48,021][lightning.pytorch.trainer.connectors.signal_connector][INFO] - Bypassing SIGTERM: 15

outputs/2026-06-04/06-24-28/.hydra/config.yaml

@@ -0,0 +1,232 @@
+experiment:
+  debug: ${debug}
+  tasks:
+  - training
+  num_nodes: 1
+  training:
+    precision: 16-mixed
+    compile: false
+    lr: 2.0e-05
+    batch_size: 16
+    max_epochs: -1
+    max_steps: 350000
+    max_time: null
+    data:
+      num_workers: 4
+      shuffle: true
+    optim:
+      accumulate_grad_batches: 1
+      gradient_clip_val: 1.0
+    checkpointing:
+      every_n_train_steps: 5000
+      every_n_epochs: null
+      train_time_interval: null
+      save_last: true
+      enable_version_counter: false
+  validation:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    val_every_n_step: 5000
+    val_every_n_epoch: null
+    limit_batch: 1
+    inference_mode: true
+    data:
+      num_workers: 4
+      shuffle: false
+  test:
+    precision: 16-mixed
+    compile: false
+    batch_size: 1
+    limit_batch: 1
+    data:
+      num_workers: 4
+      shuffle: false
+  logging:
+    metrics: null
+dataset:
+  debug: ${debug}
+  metadata: data/${dataset.name}/metadata.json
+  data_mean: 0.5
+  data_std: 0.5
+  save_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+  n_frames: 1000
+  context_length: 1
+  resolution: 128
+  observation_shape:
+  - 3
+  - ${dataset.resolution}
+  - ${dataset.resolution}
+  external_cond_dim: 0
+  validation_multiplier: 1
+  frame_skip: 1
+  action_cond_dim: 25
+  _name: video_minecraft_latent
+  precomputed_feature_dir: /proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+  n_frames_valid: 500
+  customized_validation: true
+  memory_condition_length: 0
+  wo_updown: false
+  angle_range: 180
+  pos_range: 8
+algorithm:
+  debug: ${debug}
+  lr: ${experiment.training.lr}
+  x_shape:
+  - 16
+  - 18
+  - 32
+  frame_stack: 1
+  frame_skip: ${dataset.frame_skip}
+  data_mean: ${dataset.data_mean}
+  data_std: ${dataset.data_std}
+  external_cond_dim: 0
+  context_frames: 100
+  weight_decay: 0.002
+  warmup_steps: 1000
+  optimizer_beta:
+  - 0.9
+  - 0.99
+  uncertainty_scale: 1
+  guidance_scale: 0.0
+  chunk_size: 1
+  scheduling_matrix: autoregressive
+  noise_level: random_all
+  causal: true
+  diffusion:
+    objective: pred_v
+    beta_schedule: sigmoid
+    schedule_fn_kwargs: {}
+    clip_noise: 20.0
+    use_snr: false
+    use_cum_snr: false
+    use_fused_snr: true
+    snr_clip: 5.0
+    cum_snr_decay: 0.96
+    timesteps: 1000
+    sampling_timesteps: 20
+    ddim_sampling_eta: 0.0
+    stabilization_level: 15
+    architecture:
+      network_size: 64
+      attn_heads: 4
+      attn_dim_head: 64
+      dim_mults:
+      - 1
+      - 2
+      - 4
+      - 8
+      resolution: ${dataset.resolution}
+      attn_resolutions:
+      - 16
+      - 32
+      - 64
+      - 128
+      use_init_temporal_attn: true
+      use_linear_attn: true
+      time_emb_type: rotary
+  n_frames: ${dataset.n_frames}
+  metadata: ${dataset.metadata}
+  action_cond_dim: 25
+  use_plucker: true
+  _name: dememwm_memory_dit
+  memory_token_cross_attention: true
+  memory_cross_attn_layers: null
+  memory_condition_length: 0
+  pose_cond_dim: 5
+  log_video: true
+  dememwm:
+    enabled: true
+    training_stage: stage_1
+    curriculum:
+      enabled: true
+      full_stage_start_step: 20000
+      freeze_vae: true
+      dit_freeze:
+        enabled: true
+      lr:
+        dememwm_modules: 4.0e-05
+        memory_adapters: 4.0e-05
+        full_dit: 1.0e-05
+    token_patch_size: 2
+    anchor:
+      enabled: true
+      anchor_indices:
+      - 0
+      - 1
+      - 2
+      - 3
+      allow_generated_as_anchor: false
+      diverse_selection: true
+      compress:
+        downsample_ratio: 3
+    dynamic:
+      enabled: true
+      exclude_latest_local_frames: 4
+      recent_frames: 4
+      conv_kernel_t: 3
+      conv_stride_t: 2
+    revisit:
+      enabled: true
+      deterministic_pose_retrieval: true
+      fov_overlap_threshold: 0.6
+      high_quality_fov_threshold: 0.7
+      plucker_weight: 0.1
+      max_frames: 2
+      fov_half_h: 52.5
+      fov_half_v: 37.5
+      fov_radius: 30.0
+      fov_yaw_samples: 25
+      fov_pitch_samples: 20
+      fov_depth_samples: 20
+      pose_preselect_topk: 64
+      plucker_grid_h: 4
+      plucker_grid_w: 4
+      plucker_focal_length: 0.35
+      compress:
+        downsample_ratio: 3
+    eval_ablation:
+      enabled: false
+      branch: A_plus_D_plus_R_normal
+    generated_history_proxy:
+      enabled: true
+      start_step: 40000
+      ramp_steps: 40000
+      max_prob: 0.25
+      noise_std: 0.25
+      dropout_prob: 0.0
+    injection:
+      dit_hidden_size: 1024
+      anchor_gate: 1.0
+      dynamic_gate: 1.0
+      revisit_gate: 1.0
+    cache:
+      enabled: true
+      device: cpu
+      keep_raw_latents: all
+      keep_compressed_records: true
+      keep_prefix_anchors: true
+      eviction_policy: none
+      no_evict: true
+      clear_between_videos: true
+      max_records: null
+      on_capacity_exceeded: warn
+    checkpoint:
+      strict_dememwm_eval_load: true
+  n_tokens: 4
+debug: false
+wandb:
+  entity: turlin
+  project: worldmem
+  mode: offline
+resume: null
+load: null
+auto_resume: true
+resume_ckpt_path: null
+name: train_dememwm_full_berzelius_8a100_bs16_global128_350k
+output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_full_berzelius_8a100_bs16_global128_350k/
+customized_load: true
+seperate_load: true
+diffusion_model_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+vae_path: /proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+only_tune_memory: false

outputs/2026-06-04/06-24-28/.hydra/hydra.yaml

@@ -0,0 +1,230 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    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:
+    - +name=train_dememwm_full_berzelius_8a100_bs16_global128_350k
+    - +output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_full_berzelius_8a100_bs16_global128_350k/
+    - auto_resume=true
+    - experiment.tasks=[training]
+    - algorithm=dememwm_memory_dit
+    - +customized_load=true
+    - +seperate_load=true
+    - +diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors
+    - +vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors
+    - +only_tune_memory=false
+    - dataset=video_minecraft_latent
+    - dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft
+    - dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features
+    - dataset.n_frames=1000
+    - +dataset.n_frames_valid=500
+    - +dataset.customized_validation=true
+    - +dataset.memory_condition_length=0
+    - +dataset.wo_updown=false
+    - +dataset.angle_range=180
+    - +dataset.pos_range=8
+    - ++algorithm.n_tokens=4
+    - algorithm.x_shape=[16,18,32]
+    - ++algorithm.context_frames=100
+    - ++algorithm.log_video=true
+    - ++algorithm.diffusion.sampling_timesteps=20
+    - ++algorithm.dememwm.generated_history_proxy.enabled=true
+    - ++algorithm.dememwm.generated_history_proxy.start_step=40000
+    - ++algorithm.dememwm.generated_history_proxy.ramp_steps=40000
+    - ++algorithm.dememwm.generated_history_proxy.max_prob=0.25
+    - ++algorithm.dememwm.generated_history_proxy.noise_std=0.25
+    - ++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0
+    - ++algorithm.dememwm.anchor.enabled=true
+    - ++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3]
+    - ++algorithm.dememwm.anchor.diverse_selection=true
+    - ++algorithm.dememwm.anchor.compress.downsample_ratio=3
+    - ++algorithm.dememwm.anchor.allow_generated_as_anchor=false
+    - ++algorithm.dememwm.dynamic.enabled=true
+    - ++algorithm.dememwm.dynamic.exclude_latest_local_frames=4
+    - ++algorithm.dememwm.dynamic.recent_frames=4
+    - ++algorithm.dememwm.revisit.enabled=true
+    - ++algorithm.dememwm.revisit.deterministic_pose_retrieval=true
+    - ++algorithm.dememwm.revisit.fov_overlap_threshold=0.60
+    - ++algorithm.dememwm.revisit.pose_preselect_topk=64
+    - ++algorithm.dememwm.revisit.fov_yaw_samples=25
+    - ++algorithm.dememwm.revisit.fov_pitch_samples=20
+    - ++algorithm.dememwm.revisit.fov_depth_samples=20
+    - ++algorithm.dememwm.revisit.plucker_weight=0.10
+    - ++algorithm.dememwm.revisit.max_frames=2
+    - ++algorithm.dememwm.revisit.compress.downsample_ratio=3
+    - ++algorithm.dememwm.cache.enabled=true
+    - ++algorithm.dememwm.cache.device=cpu
+    - ++algorithm.dememwm.cache.keep_raw_latents=all
+    - ++algorithm.dememwm.cache.keep_compressed_records=true
+    - ++algorithm.dememwm.cache.eviction_policy=none
+    - ++algorithm.dememwm.cache.no_evict=true
+    - ++algorithm.dememwm.cache.clear_between_videos=true
+    - ++algorithm.dememwm.cache.max_records=null
+    - ++algorithm.dememwm.cache.on_capacity_exceeded=warn
+    - ++algorithm.dememwm.curriculum.enabled=true
+    - ++algorithm.dememwm.curriculum.full_stage_start_step=20000
+    - ++algorithm.dememwm.curriculum.freeze_vae=true
+    - ++algorithm.dememwm.curriculum.dit_freeze.enabled=true
+    - ++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5
+    - ++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5
+    - experiment.training.batch_size=16
+    - experiment.training.optim.accumulate_grad_batches=1
+    - experiment.validation.batch_size=1
+    - experiment.validation.limit_batch=1
+    - experiment.training.checkpointing.every_n_train_steps=5000
+    - experiment.validation.val_every_n_step=5000
+    - experiment.training.max_steps=350000
+  job:
+    name: main
+    chdir: null
+    override_dirname: ++algorithm.context_frames=100,++algorithm.dememwm.anchor.allow_generated_as_anchor=false,++algorithm.dememwm.anchor.anchor_indices=[0,1,2,3],++algorithm.dememwm.anchor.compress.downsample_ratio=3,++algorithm.dememwm.anchor.diverse_selection=true,++algorithm.dememwm.anchor.enabled=true,++algorithm.dememwm.cache.clear_between_videos=true,++algorithm.dememwm.cache.device=cpu,++algorithm.dememwm.cache.enabled=true,++algorithm.dememwm.cache.eviction_policy=none,++algorithm.dememwm.cache.keep_compressed_records=true,++algorithm.dememwm.cache.keep_raw_latents=all,++algorithm.dememwm.cache.max_records=null,++algorithm.dememwm.cache.no_evict=true,++algorithm.dememwm.cache.on_capacity_exceeded=warn,++algorithm.dememwm.curriculum.dit_freeze.enabled=true,++algorithm.dememwm.curriculum.enabled=true,++algorithm.dememwm.curriculum.freeze_vae=true,++algorithm.dememwm.curriculum.full_stage_start_step=20000,++algorithm.dememwm.curriculum.lr.dememwm_modules=4.0e-5,++algorithm.dememwm.curriculum.lr.full_dit=1.0e-5,++algorithm.dememwm.curriculum.lr.memory_adapters=4.0e-5,++algorithm.dememwm.dynamic.enabled=true,++algorithm.dememwm.dynamic.exclude_latest_local_frames=4,++algorithm.dememwm.dynamic.recent_frames=4,++algorithm.dememwm.generated_history_proxy.dropout_prob=0.0,++algorithm.dememwm.generated_history_proxy.enabled=true,++algorithm.dememwm.generated_history_proxy.max_prob=0.25,++algorithm.dememwm.generated_history_proxy.noise_std=0.25,++algorithm.dememwm.generated_history_proxy.ramp_steps=40000,++algorithm.dememwm.generated_history_proxy.start_step=40000,++algorithm.dememwm.revisit.compress.downsample_ratio=3,++algorithm.dememwm.revisit.deterministic_pose_retrieval=true,++algorithm.dememwm.revisit.enabled=true,++algorithm.dememwm.revisit.fov_depth_samples=20,++algorithm.dememwm.revisit.fov_overlap_threshold=0.60,++algorithm.dememwm.revisit.fov_pitch_samples=20,++algorithm.dememwm.revisit.fov_yaw_samples=25,++algorithm.dememwm.revisit.max_frames=2,++algorithm.dememwm.revisit.plucker_weight=0.10,++algorithm.dememwm.revisit.pose_preselect_topk=64,++algorithm.diffusion.sampling_timesteps=20,++algorithm.log_video=true,++algorithm.n_tokens=4,+customized_load=true,+dataset.angle_range=180,+dataset.customized_validation=true,+dataset.memory_condition_length=0,+dataset.n_frames_valid=500,+dataset.pos_range=8,+dataset.wo_updown=false,+diffusion_model_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/oasis500m.safetensors,+name=train_dememwm_full_berzelius_8a100_bs16_global128_350k,+only_tune_memory=false,+output_dir=/proj/cvl/users/x_fahkh2/DeMemWM/results/dememwm_full_berzelius_8a100_bs16_global128_350k/,+seperate_load=true,+vae_path=/proj/cvl/users/x_fahkh2/WorldMem_Repro/checkpoints/vit-l-20.safetensors,algorithm.x_shape=[16,18,32],algorithm=dememwm_memory_dit,auto_resume=true,dataset.n_frames=1000,dataset.precomputed_feature_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft/vae_features,dataset.save_dir=/proj/cvl/users/x_fahkh2/WorldMem_Repro/datasets/minecraft,dataset=video_minecraft_latent,experiment.tasks=[training],experiment.training.batch_size=16,experiment.training.checkpointing.every_n_train_steps=5000,experiment.training.max_steps=350000,experiment.training.optim.accumulate_grad_batches=1,experiment.validation.batch_size=1,experiment.validation.limit_batch=1,experiment.validation.val_every_n_step=5000
+    id: ???
+    num: ???
+    config_name: training
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /proj/cvl/users/x_fahkh2/DeMemWM
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /proj/cvl/users/x_fahkh2/DeMemWM/configurations
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /proj/cvl/users/x_fahkh2/DeMemWM/outputs/2026-06-04/06-24-28
+    choices:
+      cluster: null
+      algorithm: dememwm_memory_dit
+      dataset: video_minecraft_latent
+      experiment: exp_video
+      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