Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+wandb/run-20260325_200057-qudlwnz5/run-qudlwnz5.wandb filter=lfs diff=lfs merge=lfs -text

.hydra/config.yaml

@@ -0,0 +1,327 @@
+data:
+  name: libero_cosmos_policy
+  type: mg
+  backend: robomimic
+  paths: []
+  task_suite_name: null
+  observations_keys:
+  - image
+  observation_source_keys:
+    image: agentview_rgb_jpeg
+  action_dim: 7
+  frame_stack: 1
+  horizon: 1
+  temporal_index_mode: clip
+  use_proprio: false
+  proprio_source: robot_states
+  image_chw:
+  - 3
+  - 224
+  - 224
+  image_value_range: zero_to_one
+  image_transport_dtype: uint8
+  duration_focus: null
+  hf:
+    repo_id: nvidia/LIBERO-Cosmos-Policy
+    repo_type: dataset
+    allow_patterns:
+    - success_only/*_regen/*.hdf5
+    local_files_only: false
+train:
+  optimizer:
+    type: adamw
+    lr: 0.0001
+    weight_decay: 1.0e-06
+  batch_size: 256
+  num_workers: 4
+  prefetch_factor: null
+  log_interval: 100
+  total_steps: 5000
+  eval_interval: 500
+  use_aug: true
+  aug_mode: default
+  use_amp: true
+  device_transfer_non_blocking: true
+  seed: 42
+  scheduler:
+    type: cosine
+    warmup_steps: 500
+    num_cycles: 0.5
+    min_lr_scale: 0.0
+  decoder_dataloader:
+    enabled: false
+    batch_size: 64
+    num_workers: 4
+    prefetch_factor: null
+    shuffle: true
+  stages:
+    stage1: true
+    stage2: false
+    stage3: false
+  stage1:
+    batch_size: 256
+    num_workers: 16
+    prefetch_factor: 2
+    log_interval: 100
+    total_steps: 20000
+    eval_interval: 500
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+  stage2:
+    batch_size: 256
+    num_workers: 4
+    prefetch_factor: null
+    log_interval: 100
+    total_steps: 5000
+    eval_interval: 500
+    latent_target: auto
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+    action_probe:
+      enabled: true
+  stage3:
+    batch_size: 256
+    num_workers: 4
+    prefetch_factor: null
+    log_interval: 100
+    total_steps: 5000
+    eval_interval: 500
+    final_eval_num_episodes: 100
+    log_eval_video: false
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+    trainable_scope: all
+    non_decoder_lr_scale: 1.0
+  checkpoint:
+    enabled: true
+    base_dir: null
+    save_interval: 1000
+    save_last: true
+    latest_only: true
+    load:
+      stage1:
+        mode: none
+        path: null
+      stage2:
+        mode: handoff
+        path: null
+        teacher_path: null
+      stage3:
+        mode: handoff
+        path: null
+model:
+  idm:
+    type: token_idm
+    action_dim: 128
+    token_dim: 768
+    model_dim: 512
+    latent_dim: 32
+    num_action_tokens: 4
+    num_blocks: 4
+    num_heads: 8
+    dropout: 0.0
+  fdm:
+    type: token_fdm_duration
+    action_dim: 128
+    token_dim: 768
+    model_dim: 512
+    num_action_tokens: 4
+    num_blocks: 4
+    num_heads: 8
+    dropout: 0.0
+  hyperbolic_latent:
+    enabled: false
+    backend: geoopt
+    manifold: poincare
+    curvature: 1.0
+    learnable_curvature: false
+    fdm_input_mode: logmap0
+    prelift_mode: none
+    prelift_scale: 1.0
+    lift_max_norm: 5.0
+    tangent_max_norm: 5.0
+    eps: 1.0e-06
+  encoders:
+    type: dino
+    image_key: image
+    model_id: facebook/dinov2-base
+    input_value_range: zero_to_one
+    freeze_backbone: true
+    drop_cls_token: true
+    output_dim: 768
+    mean:
+    - 0.485
+    - 0.456
+    - 0.406
+    std:
+    - 0.229
+    - 0.224
+    - 0.225
+  type: dino_lam
+  idm_input: future
+  fdm_target: future
+  pixel_decoders: null
+  latent_action_decoders: null
+objective:
+  fdm_target: future
+  idm_input: future
+  stage_overrides: {}
+  multiscale:
+    enabled: true
+    consistency:
+      enabled: false
+      weight: 10.0
+      num_pairs: 4
+      sample_source: all_horizons
+      teacher_mode: direct_teacher
+    prediction_mode: direct_duration
+    allow_plain_fdm_oneshot: true
+    weight_mode: uniform
+    weights: {}
+  temporal:
+    max_offset: 20
+    anchor_mode: fixed
+    extra_random_count: 0
+  branch_order:
+    enabled: false
+    radial_weight: 0.0
+    local_radial_margin_weight: 0.0
+    local_radial_margin_alpha: 0.05
+    branch_weight: 0.0
+    z0_origin_weight: 0.0
+    prefix_weight: 0.0
+    radius_progress_weight: 1.0
+    radius_progress_mode: offset_margin
+    radius_progress_alpha: 0.02
+    branch_margin_deg: 10.0
+    eps: 1.0e-06
+  latent_plan:
+    enabled: false
+    total_horizon: 20
+eval:
+  type: robomimic
+  data_path: ${data.paths}
+  reset_mode: env_reset
+  num_eval_episodes: 20
+  max_steps: 500
+  record_video: true
+  checkpoint_path: null
+  checkpoint_strict: true
+  use_checkpoint_cfg: true
+  video_output_path: null
+  obs_keys: null
+  use_proprio: null
+  use_object: false
+  resize_hw: null
+  image_value_range: null
+  seed: 0
+agent:
+  encoders:
+    image:
+      in_channels: 3
+      output_dim: 512
+      output_mode: global
+      type: resnet18
+      pretrained: false
+    proprio:
+      input_dim: 9
+      hidden_dim: 128
+      output_dim: 64
+    type: group
+    modalities:
+    - agentview_image
+    - robot0_eye_in_hand_image
+    proj_dim: 128
+  policies:
+    decoder:
+      type: mlp
+      hidden_dims:
+      - 256
+      - 256
+    hidden_dims:
+    - 512
+    - 512
+    - 256
+    - 64
+    action_dim: 256
+    emb_dim: 384
+    gt_action_dim: 7
+  type: latent_action
+probes:
+  enabled: true
+  every: 10
+  steps_per_call: 1
+  sequence:
+    enabled: true
+    horizons:
+    - 20
+    target_key: gt_action_seq_max
+  list:
+    z_to_s_t:
+      name: z_to_s_t
+      enabled: false
+      shuffle: true
+      type: regression
+      input: z_t
+      target: s_t
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_s_tp:
+      name: z_to_s_tp
+      enabled: false
+      shuffle: true
+      type: regression
+      input: z_t
+      target: s_tp
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_action_t:
+      name: z_to_action_t
+      probe_type: z_to_action
+      enabled: true
+      shuffle: true
+      type: regression
+      input: z_t
+      target: gt_action
+      loss: mse
+      lr: 0.001
+      every: 1
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_action_seq_h20:
+      name: z_to_action_seq_h20
+      probe_type: z_to_action_sequence
+      enabled: true
+      shuffle: true
+      type: regression
+      input: z_t
+      target: gt_action_seq_max
+      sequence_horizon: 20
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+logger:
+  project: latent_action
+  run_name: latent_action_training
+  tags:
+  - debug
+output_root: outputs/token_fdm_duration/dino_suite_all_k20_extra0

.hydra/hydra.yaml

@@ -0,0 +1,192 @@
+hydra:
+  run:
+    dir: ${output_root}/${data.name}/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  sweep:
+    dir: ${output_root}/multirun/${data.name}/${now:%Y-%m-%d}/${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:
+    - data=libero_cosmos_policy
+    - data.task_suite_name=null
+    - data.hf.allow_patterns=[success_only/*_regen/*.hdf5]
+    - data.image_chw=[3,224,224]
+    - data.temporal_index_mode=clip
+    - model.fdm.type=token_fdm_duration
+    - objective.idm_input=future
+    - objective.temporal.max_offset=20
+    - objective.temporal.anchor_mode=fixed
+    - objective.temporal.extra_random_count=0
+    - train.use_aug=true
+    - train.use_amp=true
+    - train.stage1.batch_size=256
+    - train.stage1.num_workers=16
+    - train.stage1.prefetch_factor=2
+    - train.stage1.total_steps=20000
+    - train.stage1.log_interval=100
+    - train.checkpoint.enabled=true
+    - train.checkpoint.save_interval=1000
+    - train.checkpoint.save_last=true
+    - train.checkpoint.latest_only=true
+    - output_root=outputs/token_fdm_duration/dino_suite_all_k20_extra0
+  job:
+    name: run
+    chdir: null
+    override_dirname: data.hf.allow_patterns=[success_only/*_regen/*.hdf5],data.image_chw=[3,224,224],data.task_suite_name=null,data.temporal_index_mode=clip,data=libero_cosmos_policy,model.fdm.type=token_fdm_duration,objective.idm_input=future,objective.temporal.anchor_mode=fixed,objective.temporal.extra_random_count=0,objective.temporal.max_offset=20,output_root=outputs/token_fdm_duration/dino_suite_all_k20_extra0,train.checkpoint.enabled=true,train.checkpoint.latest_only=true,train.checkpoint.save_interval=1000,train.checkpoint.save_last=true,train.stage1.batch_size=256,train.stage1.log_interval=100,train.stage1.num_workers=16,train.stage1.prefetch_factor=2,train.stage1.total_steps=20000,train.use_amp=true,train.use_aug=true
+    id: ???
+    num: ???
+    config_name: dino_lam
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.1'
+    cwd: /home/jovyan/youngjoonjeong/latent_action
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /home/jovyan/youngjoonjeong/latent_action/configs
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52
+    choices:
+      logger: wandb
+      probes: action_t
+      agent: latent_action_agent
+      model/policies@agent.policies: latent_action
+      model/policies/decoder@agent.policies.decoder: mlp
+      model/encoders@agent.encoders: group_encoder
+      model/encoders/proprio@agent.encoders.proprio: mlp
+      model/encoders/image@agent.encoders.image: resnet18
+      eval: robomimic_eval
+      objective: multiscale
+      model: dino_lam
+      model/encoders: dino_encoder
+      model/hyperbolic_latent: base
+      train: latent_action_trainer
+      train/optimizer: adamw
+      data: libero_cosmos_policy
+      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

.hydra/overrides.yaml

@@ -0,0 +1,22 @@
+- data=libero_cosmos_policy
+- data.task_suite_name=null
+- data.hf.allow_patterns=[success_only/*_regen/*.hdf5]
+- data.image_chw=[3,224,224]
+- data.temporal_index_mode=clip
+- model.fdm.type=token_fdm_duration
+- objective.idm_input=future
+- objective.temporal.max_offset=20
+- objective.temporal.anchor_mode=fixed
+- objective.temporal.extra_random_count=0
+- train.use_aug=true
+- train.use_amp=true
+- train.stage1.batch_size=256
+- train.stage1.num_workers=16
+- train.stage1.prefetch_factor=2
+- train.stage1.total_steps=20000
+- train.stage1.log_interval=100
+- train.checkpoint.enabled=true
+- train.checkpoint.save_interval=1000
+- train.checkpoint.save_last=true
+- train.checkpoint.latest_only=true
+- output_root=outputs/token_fdm_duration/dino_suite_all_k20_extra0

config_resolved.yaml

@@ -0,0 +1,333 @@
+data:
+  name: libero_cosmos_policy
+  type: mg
+  backend: robomimic
+  paths: []
+  task_suite_name: null
+  observations_keys:
+  - image
+  observation_source_keys:
+    image: agentview_rgb_jpeg
+  action_dim: 7
+  frame_stack: 1
+  horizon: 1
+  temporal_index_mode: clip
+  use_proprio: false
+  proprio_source: robot_states
+  image_chw:
+  - 3
+  - 224
+  - 224
+  image_value_range: zero_to_one
+  image_transport_dtype: uint8
+  duration_focus: null
+  hf:
+    repo_id: nvidia/LIBERO-Cosmos-Policy
+    repo_type: dataset
+    allow_patterns:
+    - success_only/*_regen/*.hdf5
+    local_files_only: false
+  action_sequence_targets:
+    enabled: true
+    horizons:
+    - 20
+    target_key: gt_action_seq_max
+train:
+  optimizer:
+    type: adamw
+    lr: 0.0001
+    weight_decay: 1.0e-06
+  batch_size: 256
+  num_workers: 4
+  prefetch_factor: null
+  log_interval: 100
+  total_steps: 5000
+  eval_interval: 500
+  use_aug: true
+  aug_mode: default
+  use_amp: true
+  device_transfer_non_blocking: true
+  seed: 42
+  scheduler:
+    type: cosine
+    warmup_steps: 500
+    num_cycles: 0.5
+    min_lr_scale: 0.0
+  decoder_dataloader:
+    enabled: false
+    batch_size: 64
+    num_workers: 4
+    prefetch_factor: null
+    shuffle: true
+  stages:
+    stage1: true
+    stage2: false
+    stage3: false
+  stage1:
+    batch_size: 256
+    num_workers: 16
+    prefetch_factor: 2
+    log_interval: 100
+    total_steps: 20000
+    eval_interval: 500
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+  stage2:
+    batch_size: 256
+    num_workers: 4
+    prefetch_factor: null
+    log_interval: 100
+    total_steps: 5000
+    eval_interval: 500
+    latent_target: auto
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+    action_probe:
+      enabled: true
+  stage3:
+    batch_size: 256
+    num_workers: 4
+    prefetch_factor: null
+    log_interval: 100
+    total_steps: 5000
+    eval_interval: 500
+    final_eval_num_episodes: 100
+    log_eval_video: false
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+    trainable_scope: all
+    non_decoder_lr_scale: 1.0
+  checkpoint:
+    enabled: true
+    base_dir: null
+    save_interval: 1000
+    save_last: true
+    latest_only: true
+    load:
+      stage1:
+        mode: none
+        path: null
+      stage2:
+        mode: handoff
+        path: null
+        teacher_path: null
+      stage3:
+        mode: handoff
+        path: null
+model:
+  idm:
+    type: token_idm
+    action_dim: 128
+    token_dim: 768
+    model_dim: 512
+    latent_dim: 32
+    num_action_tokens: 4
+    num_blocks: 4
+    num_heads: 8
+    dropout: 0.0
+  fdm:
+    type: token_fdm_duration
+    action_dim: 128
+    token_dim: 768
+    model_dim: 512
+    num_action_tokens: 4
+    num_blocks: 4
+    num_heads: 8
+    dropout: 0.0
+  hyperbolic_latent:
+    enabled: false
+    backend: geoopt
+    manifold: poincare
+    curvature: 1.0
+    learnable_curvature: false
+    fdm_input_mode: logmap0
+    prelift_mode: none
+    prelift_scale: 1.0
+    lift_max_norm: 5.0
+    tangent_max_norm: 5.0
+    eps: 1.0e-06
+  encoders:
+    type: dino
+    image_key: image
+    model_id: facebook/dinov2-base
+    input_value_range: zero_to_one
+    freeze_backbone: true
+    drop_cls_token: true
+    output_dim: 768
+    mean:
+    - 0.485
+    - 0.456
+    - 0.406
+    std:
+    - 0.229
+    - 0.224
+    - 0.225
+  type: dino_lam
+  idm_input: future
+  fdm_target: future
+  pixel_decoders: null
+  latent_action_decoders: null
+objective:
+  fdm_target: future
+  idm_input: future
+  stage_overrides: {}
+  multiscale:
+    enabled: true
+    consistency:
+      enabled: false
+      weight: 10.0
+      num_pairs: 4
+      sample_source: all_horizons
+      teacher_mode: direct_teacher
+    prediction_mode: direct_duration
+    allow_plain_fdm_oneshot: true
+    weight_mode: uniform
+    weights: {}
+  temporal:
+    max_offset: 20
+    anchor_mode: fixed
+    extra_random_count: 0
+  branch_order:
+    enabled: false
+    radial_weight: 0.0
+    local_radial_margin_weight: 0.0
+    local_radial_margin_alpha: 0.05
+    branch_weight: 0.0
+    z0_origin_weight: 0.0
+    prefix_weight: 0.0
+    radius_progress_weight: 1.0
+    radius_progress_mode: offset_margin
+    radius_progress_alpha: 0.02
+    branch_margin_deg: 10.0
+    eps: 1.0e-06
+  latent_plan:
+    enabled: false
+    total_horizon: 20
+eval:
+  type: robomimic
+  data_path: []
+  reset_mode: env_reset
+  num_eval_episodes: 20
+  max_steps: 500
+  record_video: true
+  checkpoint_path: null
+  checkpoint_strict: true
+  use_checkpoint_cfg: true
+  video_output_path: null
+  obs_keys: null
+  use_proprio: null
+  use_object: false
+  resize_hw: null
+  image_value_range: null
+  seed: 0
+agent:
+  encoders:
+    image:
+      in_channels: 3
+      output_dim: 512
+      output_mode: global
+      type: resnet18
+      pretrained: false
+    proprio:
+      input_dim: 9
+      hidden_dim: 128
+      output_dim: 64
+    type: group
+    modalities:
+    - agentview_image
+    - robot0_eye_in_hand_image
+    proj_dim: 128
+  policies:
+    decoder:
+      type: mlp
+      hidden_dims:
+      - 256
+      - 256
+    hidden_dims:
+    - 512
+    - 512
+    - 256
+    - 64
+    action_dim: 256
+    emb_dim: 384
+    gt_action_dim: 7
+  type: latent_action
+probes:
+  enabled: true
+  every: 10
+  steps_per_call: 1
+  sequence:
+    enabled: true
+    horizons:
+    - 20
+    target_key: gt_action_seq_max
+  list:
+    z_to_s_t:
+      name: z_to_s_t
+      enabled: false
+      shuffle: true
+      type: regression
+      input: z_t
+      target: s_t
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_s_tp:
+      name: z_to_s_tp
+      enabled: false
+      shuffle: true
+      type: regression
+      input: z_t
+      target: s_tp
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_action_t:
+      name: z_to_action_t
+      probe_type: z_to_action
+      enabled: true
+      shuffle: true
+      type: regression
+      input: z_t
+      target: gt_action
+      loss: mse
+      lr: 0.001
+      every: 1
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_action_seq_h20:
+      name: z_to_action_seq_h20
+      probe_type: z_to_action_sequence
+      enabled: true
+      shuffle: true
+      type: regression
+      input: z_t
+      target: gt_action_seq_max
+      sequence_horizon: 20
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+      output_dim: 140
+logger:
+  project: latent_action
+  run_name: latent_action_training
+  tags:
+  - debug
+output_root: outputs/token_fdm_duration/dino_suite_all_k20_extra0

git.json

@@ -0,0 +1,9 @@
+{
+  "branch": "integration/master-hyperbolic",
+  "captured_at": "2026-03-25T20:00:54.494786+00:00",
+  "commit": "84bdcca892435da456b8a676dae358493a0a6e59",
+  "dirty": false,
+  "repo_root": "/home/jovyan/youngjoonjeong/latent_action",
+  "short_commit": "84bdcca",
+  "status": []
+}

hydra_resolved.yaml

@@ -0,0 +1,333 @@
+data:
+  name: libero_cosmos_policy
+  type: mg
+  backend: robomimic
+  paths: []
+  task_suite_name: null
+  observations_keys:
+  - image
+  observation_source_keys:
+    image: agentview_rgb_jpeg
+  action_dim: 7
+  frame_stack: 1
+  horizon: 1
+  temporal_index_mode: clip
+  use_proprio: false
+  proprio_source: robot_states
+  image_chw:
+  - 3
+  - 224
+  - 224
+  image_value_range: zero_to_one
+  image_transport_dtype: uint8
+  duration_focus: null
+  hf:
+    repo_id: nvidia/LIBERO-Cosmos-Policy
+    repo_type: dataset
+    allow_patterns:
+    - success_only/*_regen/*.hdf5
+    local_files_only: false
+  action_sequence_targets:
+    enabled: true
+    horizons:
+    - 20
+    target_key: gt_action_seq_max
+train:
+  optimizer:
+    type: adamw
+    lr: 0.0001
+    weight_decay: 1.0e-06
+  batch_size: 256
+  num_workers: 4
+  prefetch_factor: null
+  log_interval: 100
+  total_steps: 5000
+  eval_interval: 500
+  use_aug: true
+  aug_mode: default
+  use_amp: true
+  device_transfer_non_blocking: true
+  seed: 42
+  scheduler:
+    type: cosine
+    warmup_steps: 500
+    num_cycles: 0.5
+    min_lr_scale: 0.0
+  decoder_dataloader:
+    enabled: false
+    batch_size: 64
+    num_workers: 4
+    prefetch_factor: null
+    shuffle: true
+  stages:
+    stage1: true
+    stage2: false
+    stage3: false
+  stage1:
+    batch_size: 256
+    num_workers: 16
+    prefetch_factor: 2
+    log_interval: 100
+    total_steps: 20000
+    eval_interval: 500
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+  stage2:
+    batch_size: 256
+    num_workers: 4
+    prefetch_factor: null
+    log_interval: 100
+    total_steps: 5000
+    eval_interval: 500
+    latent_target: auto
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+    action_probe:
+      enabled: true
+  stage3:
+    batch_size: 256
+    num_workers: 4
+    prefetch_factor: null
+    log_interval: 100
+    total_steps: 5000
+    eval_interval: 500
+    final_eval_num_episodes: 100
+    log_eval_video: false
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+    trainable_scope: all
+    non_decoder_lr_scale: 1.0
+  checkpoint:
+    enabled: true
+    base_dir: null
+    save_interval: 1000
+    save_last: true
+    latest_only: true
+    load:
+      stage1:
+        mode: none
+        path: null
+      stage2:
+        mode: handoff
+        path: null
+        teacher_path: null
+      stage3:
+        mode: handoff
+        path: null
+model:
+  idm:
+    type: token_idm
+    action_dim: 128
+    token_dim: 768
+    model_dim: 512
+    latent_dim: 32
+    num_action_tokens: 4
+    num_blocks: 4
+    num_heads: 8
+    dropout: 0.0
+  fdm:
+    type: token_fdm_duration
+    action_dim: 128
+    token_dim: 768
+    model_dim: 512
+    num_action_tokens: 4
+    num_blocks: 4
+    num_heads: 8
+    dropout: 0.0
+  hyperbolic_latent:
+    enabled: false
+    backend: geoopt
+    manifold: poincare
+    curvature: 1.0
+    learnable_curvature: false
+    fdm_input_mode: logmap0
+    prelift_mode: none
+    prelift_scale: 1.0
+    lift_max_norm: 5.0
+    tangent_max_norm: 5.0
+    eps: 1.0e-06
+  encoders:
+    type: dino
+    image_key: image
+    model_id: facebook/dinov2-base
+    input_value_range: zero_to_one
+    freeze_backbone: true
+    drop_cls_token: true
+    output_dim: 768
+    mean:
+    - 0.485
+    - 0.456
+    - 0.406
+    std:
+    - 0.229
+    - 0.224
+    - 0.225
+  type: dino_lam
+  idm_input: future
+  fdm_target: future
+  pixel_decoders: null
+  latent_action_decoders: null
+objective:
+  fdm_target: future
+  idm_input: future
+  stage_overrides: {}
+  multiscale:
+    enabled: true
+    consistency:
+      enabled: false
+      weight: 10.0
+      num_pairs: 4
+      sample_source: all_horizons
+      teacher_mode: direct_teacher
+    prediction_mode: direct_duration
+    allow_plain_fdm_oneshot: true
+    weight_mode: uniform
+    weights: {}
+  temporal:
+    max_offset: 20
+    anchor_mode: fixed
+    extra_random_count: 0
+  branch_order:
+    enabled: false
+    radial_weight: 0.0
+    local_radial_margin_weight: 0.0
+    local_radial_margin_alpha: 0.05
+    branch_weight: 0.0
+    z0_origin_weight: 0.0
+    prefix_weight: 0.0
+    radius_progress_weight: 1.0
+    radius_progress_mode: offset_margin
+    radius_progress_alpha: 0.02
+    branch_margin_deg: 10.0
+    eps: 1.0e-06
+  latent_plan:
+    enabled: false
+    total_horizon: 20
+eval:
+  type: robomimic
+  data_path: []
+  reset_mode: env_reset
+  num_eval_episodes: 20
+  max_steps: 500
+  record_video: true
+  checkpoint_path: null
+  checkpoint_strict: true
+  use_checkpoint_cfg: true
+  video_output_path: null
+  obs_keys: null
+  use_proprio: null
+  use_object: false
+  resize_hw: null
+  image_value_range: null
+  seed: 0
+agent:
+  encoders:
+    image:
+      in_channels: 3
+      output_dim: 512
+      output_mode: global
+      type: resnet18
+      pretrained: false
+    proprio:
+      input_dim: 9
+      hidden_dim: 128
+      output_dim: 64
+    type: group
+    modalities:
+    - agentview_image
+    - robot0_eye_in_hand_image
+    proj_dim: 128
+  policies:
+    decoder:
+      type: mlp
+      hidden_dims:
+      - 256
+      - 256
+    hidden_dims:
+    - 512
+    - 512
+    - 256
+    - 64
+    action_dim: 256
+    emb_dim: 384
+    gt_action_dim: 7
+  type: latent_action
+probes:
+  enabled: true
+  every: 10
+  steps_per_call: 1
+  sequence:
+    enabled: true
+    horizons:
+    - 20
+    target_key: gt_action_seq_max
+  list:
+    z_to_s_t:
+      name: z_to_s_t
+      enabled: false
+      shuffle: true
+      type: regression
+      input: z_t
+      target: s_t
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_s_tp:
+      name: z_to_s_tp
+      enabled: false
+      shuffle: true
+      type: regression
+      input: z_t
+      target: s_tp
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_action_t:
+      name: z_to_action_t
+      probe_type: z_to_action
+      enabled: true
+      shuffle: true
+      type: regression
+      input: z_t
+      target: gt_action
+      loss: mse
+      lr: 0.001
+      every: 1
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_action_seq_h20:
+      name: z_to_action_seq_h20
+      probe_type: z_to_action_sequence
+      enabled: true
+      shuffle: true
+      type: regression
+      input: z_t
+      target: gt_action_seq_max
+      sequence_horizon: 20
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+      output_dim: 140
+logger:
+  project: latent_action
+  run_name: latent_action_training
+  tags:
+  - debug
+output_root: outputs/token_fdm_duration/dino_suite_all_k20_extra0

intent.md

@@ -0,0 +1,13 @@
+# Intent
+
+Hypothesis:
+- 
+
+Primary comparison:
+- 
+
+Success criterion:
+- 
+
+Failure modes to watch:
+- 

notes.md

@@ -0,0 +1,10 @@
+# Notes
+
+Result summary:
+- 
+
+Interpretation:
+- 
+
+Next action:
+- 

outputs/token_fdm_duration/dino_suite_all_k20_extra0/registry.jsonl

@@ -0,0 +1 @@
+{"base_run_name": "latent_action_training", "config_resolved_yaml": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/config_resolved.yaml", "created_at": "2026-03-25T20:00:54.495122+00:00", "data_name": "libero_cosmos_policy", "enabled_stages": {"stage1": true, "stage2": false, "stage3": false}, "git_json": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/git.json", "intent_md": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/intent.md", "notes_md": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/notes.md", "run_dir": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52"}

registry_entry.json

@@ -0,0 +1,15 @@
+{
+  "base_run_name": "latent_action_training",
+  "config_resolved_yaml": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/config_resolved.yaml",
+  "created_at": "2026-03-25T20:00:54.495122+00:00",
+  "data_name": "libero_cosmos_policy",
+  "enabled_stages": {
+    "stage1": true,
+    "stage2": false,
+    "stage3": false
+  },
+  "git_json": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/git.json",
+  "intent_md": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/intent.md",
+  "notes_md": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/notes.md",
+  "run_dir": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52"
+}

stages/stage1/checkpoint_latest.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3987875cf73ed546f1d6d4568bacc54c695ab49ade5a4735caf5528195989f8c
+size 671070930

stages/stage1/hydra_resolved.yaml

@@ -0,0 +1,423 @@
+data:
+  name: libero_cosmos_policy
+  type: mg
+  backend: robomimic
+  paths:
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE3_turn_on_the_stove_and_put_the_moka_pot_on_it_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE4_put_the_black_bowl_in_the_bottom_drawer_of_the_cabinet_and_close_it_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE6_put_the_yellow_and_white_mug_in_the_microwave_and_close_it_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE8_put_both_moka_pots_on_the_stove_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE1_put_both_the_alphabet_soup_and_the_cream_cheese_box_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE2_put_both_the_alphabet_soup_and_the_tomato_sauce_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE2_put_both_the_cream_cheese_box_and_the_butter_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE5_put_the_white_mug_on_the_left_plate_and_put_the_yellow_and_white_mug_on_the_right_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE6_put_the_white_mug_on_the_plate_and_put_the_chocolate_pudding_to_the_right_of_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/STUDY_SCENE1_pick_up_the_book_and_place_it_in_the_back_compartment_of_the_caddy_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/open_the_middle_drawer_of_the_cabinet_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/open_the_top_drawer_and_put_the_bowl_inside_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/push_the_plate_to_the_front_of_the_stove_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_the_stove_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_top_of_the_cabinet_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_cream_cheese_in_the_bowl_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_wine_bottle_on_the_rack_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_wine_bottle_on_top_of_the_cabinet_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/turn_on_the_stove_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_alphabet_soup_and_place_it_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_bbq_sauce_and_place_it_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_butter_and_place_it_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_chocolate_pudding_and_place_it_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_cream_cheese_and_place_it_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_ketchup_and_place_it_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_milk_and_place_it_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_orange_juice_and_place_it_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_salad_dressing_and_place_it_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_tomato_sauce_and_place_it_in_the_basket_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_between_the_plate_and_the_ramekin_and_place_it_on_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_from_table_center_and_place_it_on_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_in_the_top_drawer_of_the_wooden_cabinet_and_place_it_on_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_cookie_box_and_place_it_on_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_plate_and_place_it_on_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_ramekin_and_place_it_on_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_cookie_box_and_place_it_on_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_ramekin_and_place_it_on_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_stove_and_place_it_on_the_plate_demo.hdf5
+  - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_wooden_cabinet_and_place_it_on_the_plate_demo.hdf5
+  task_suite_name: null
+  observations_keys:
+  - image
+  observation_source_keys:
+    image: agentview_rgb_jpeg
+  action_dim: 7
+  frame_stack: 1
+  horizon: 1
+  temporal_index_mode: clip
+  use_proprio: false
+  proprio_source: robot_states
+  image_chw:
+  - 3
+  - 224
+  - 224
+  image_value_range: zero_to_one
+  image_transport_dtype: uint8
+  duration_focus: null
+  hf:
+    repo_id: nvidia/LIBERO-Cosmos-Policy
+    repo_type: dataset
+    allow_patterns:
+    - success_only/*_regen/*.hdf5
+    local_files_only: false
+  action_sequence_targets:
+    enabled: true
+    horizons:
+    - 20
+    target_key: gt_action_seq_max
+  temporal:
+    max_offset: 20
+    anchor_mode: fixed
+    extra_random_count: 0
+    load_full_context: false
+  multiscale:
+    enabled: true
+  use_aug: false
+  emit_states: false
+  emit_duration_label: false
+  emit_future_offset_fields: false
+  sample_transform:
+    weak_augmentation:
+      enabled: false
+train:
+  optimizer:
+    type: adamw
+    lr: 0.0001
+    weight_decay: 1.0e-06
+  batch_size: 256
+  num_workers: 4
+  prefetch_factor: null
+  log_interval: 100
+  total_steps: 5000
+  eval_interval: 500
+  use_aug: true
+  aug_mode: default
+  use_amp: true
+  device_transfer_non_blocking: true
+  seed: 42
+  scheduler:
+    type: cosine
+    warmup_steps: 500
+    num_cycles: 0.5
+    min_lr_scale: 0.0
+  decoder_dataloader:
+    enabled: false
+    batch_size: 64
+    num_workers: 4
+    prefetch_factor: null
+    shuffle: true
+  stages:
+    stage1: true
+    stage2: false
+    stage3: false
+  stage1:
+    batch_size: 256
+    num_workers: 16
+    prefetch_factor: 2
+    log_interval: 100
+    total_steps: 20000
+    eval_interval: 500
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+  stage2:
+    batch_size: 256
+    num_workers: 4
+    prefetch_factor: null
+    log_interval: 100
+    total_steps: 5000
+    eval_interval: 500
+    latent_target: auto
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+    action_probe:
+      enabled: true
+  stage3:
+    batch_size: 256
+    num_workers: 4
+    prefetch_factor: null
+    log_interval: 100
+    total_steps: 5000
+    eval_interval: 500
+    final_eval_num_episodes: 100
+    log_eval_video: false
+    optimizer:
+      lr: 0.0001
+      weight_decay: 1.0e-06
+    trainable_scope: all
+    non_decoder_lr_scale: 1.0
+  checkpoint:
+    enabled: true
+    base_dir: null
+    save_interval: 1000
+    save_last: true
+    latest_only: true
+    load:
+      stage1:
+        mode: none
+        path: null
+      stage2:
+        mode: handoff
+        path: null
+        teacher_path: null
+      stage3:
+        mode: handoff
+        path: null
+model:
+  idm:
+    type: token_idm
+    action_dim: 128
+    token_dim: 768
+    model_dim: 512
+    latent_dim: 32
+    num_action_tokens: 4
+    num_blocks: 4
+    num_heads: 8
+    dropout: 0.0
+  fdm:
+    type: token_fdm_duration
+    action_dim: 128
+    token_dim: 768
+    model_dim: 512
+    num_action_tokens: 4
+    num_blocks: 4
+    num_heads: 8
+    dropout: 0.0
+    prediction_mode: direct
+  hyperbolic_latent:
+    enabled: false
+    backend: geoopt
+    manifold: poincare
+    curvature: 1.0
+    learnable_curvature: false
+    fdm_input_mode: logmap0
+    prelift_mode: none
+    prelift_scale: 1.0
+    lift_max_norm: 5.0
+    tangent_max_norm: 5.0
+    eps: 1.0e-06
+  encoders:
+    type: dino
+    image_key: image
+    model_id: facebook/dinov2-base
+    input_value_range: zero_to_one
+    freeze_backbone: true
+    drop_cls_token: true
+    output_dim: 768
+    mean:
+    - 0.485
+    - 0.456
+    - 0.406
+    std:
+    - 0.229
+    - 0.224
+    - 0.225
+  type: dino_lam
+  idm_input: future
+  fdm_target: future
+  pixel_decoders: null
+  latent_action_decoders: null
+  temporal:
+    max_offset: 20
+    anchor_mode: fixed
+    extra_random_count: 0
+    load_full_context: false
+  multiscale_loss:
+    enabled: true
+    prediction_mode: direct_duration
+    weight_mode: uniform
+    weights: {}
+    allow_plain_fdm_oneshot: true
+    direct_duration_backend: duration_aware
+    consistency:
+      enabled: false
+      weight: 10.0
+      num_pairs: 4
+      sample_source: all_horizons
+      teacher_mode: direct_teacher
+  branch_order:
+    enabled: false
+    radial_weight: 0.0
+    local_radial_margin_weight: 0.0
+    local_radial_margin_alpha: 0.05
+    branch_weight: 0.0
+    z0_origin_weight: 0.0
+    prefix_weight: 0.0
+    radius_progress_weight: 1.0
+    radius_progress_mode: offset_margin
+    radius_progress_alpha: 0.02
+    branch_margin_deg: 10.0
+    branch_margin_rad: 0.17453292519943295
+    eps: 1.0e-06
+objective:
+  fdm_target: future
+  idm_input: future
+  multiscale:
+    enabled: true
+    consistency:
+      enabled: false
+      weight: 10.0
+      num_pairs: 4
+      sample_source: all_horizons
+      teacher_mode: direct_teacher
+    prediction_mode: direct_duration
+    allow_plain_fdm_oneshot: true
+    weight_mode: uniform
+    weights: {}
+  temporal:
+    max_offset: 20
+    anchor_mode: fixed
+    extra_random_count: 0
+    load_full_context: false
+  branch_order:
+    enabled: false
+    radial_weight: 0.0
+    local_radial_margin_weight: 0.0
+    local_radial_margin_alpha: 0.05
+    branch_weight: 0.0
+    z0_origin_weight: 0.0
+    prefix_weight: 0.0
+    radius_progress_weight: 1.0
+    radius_progress_mode: offset_margin
+    radius_progress_alpha: 0.02
+    branch_margin_deg: 10.0
+    eps: 1.0e-06
+  latent_plan:
+    enabled: false
+    total_horizon: 20
+    num_chunks: 1
+    chunk_dim: 128
+    plan_dim: 128
+eval:
+  type: robomimic
+  data_path: []
+  reset_mode: env_reset
+  num_eval_episodes: 20
+  max_steps: 500
+  record_video: true
+  checkpoint_path: null
+  checkpoint_strict: true
+  use_checkpoint_cfg: true
+  video_output_path: null
+  obs_keys: null
+  use_proprio: null
+  use_object: false
+  resize_hw: null
+  image_value_range: null
+  seed: 0
+agent:
+  encoders:
+    image:
+      in_channels: 3
+      output_dim: 512
+      output_mode: global
+      type: resnet18
+      pretrained: false
+    proprio:
+      input_dim: 9
+      hidden_dim: 128
+      output_dim: 64
+    type: group
+    modalities:
+    - agentview_image
+    - robot0_eye_in_hand_image
+    proj_dim: 128
+  policies:
+    decoder:
+      type: mlp
+      hidden_dims:
+      - 256
+      - 256
+    hidden_dims:
+    - 512
+    - 512
+    - 256
+    - 64
+    action_dim: 256
+    emb_dim: 384
+    gt_action_dim: 7
+  type: latent_action
+probes:
+  enabled: true
+  every: 10
+  steps_per_call: 1
+  sequence:
+    enabled: true
+    horizons:
+    - 20
+    target_key: gt_action_seq_max
+  list:
+    z_to_s_t:
+      name: z_to_s_t
+      enabled: false
+      shuffle: true
+      type: regression
+      input: z_t
+      target: s_t
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_s_tp:
+      name: z_to_s_tp
+      enabled: false
+      shuffle: true
+      type: regression
+      input: z_t
+      target: s_tp
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_action_t:
+      name: z_to_action_t
+      probe_type: z_to_action
+      enabled: true
+      shuffle: true
+      type: regression
+      input: z_t
+      target: gt_action
+      loss: mse
+      lr: 0.001
+      every: 1
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+    z_to_action_seq_h20:
+      name: z_to_action_seq_h20
+      probe_type: z_to_action_sequence
+      enabled: true
+      shuffle: true
+      type: regression
+      input: z_t
+      target: gt_action_seq_max
+      sequence_horizon: 20
+      loss: mse
+      lr: 0.001
+      every: 10
+      mlp:
+        hidden_dims:
+        - 128
+        - 64
+      output_dim: 140
+logger:
+  project: latent_action
+  run_name: latent_action_training
+  tags:
+  - debug
+output_root: outputs/token_fdm_duration/dino_suite_all_k20_extra0

stages/stage1/parent_run.json

@@ -0,0 +1,7 @@
+{
+  "explicit_path": null,
+  "mode": "none",
+  "slurm_job_id": null,
+  "source_stage": null,
+  "stage": "stage1"
+}

stages/stage1/summary.json

@@ -0,0 +1,65 @@
+{
+  "best_metrics": {
+    "loss/fdm_fit_loss": 0.2659306824207306,
+    "loss/fdm_h20": 0.2659306824207306,
+    "loss/total_loss": 0.2659306824207306,
+    "probe/z_to_action_seq_h20/l2": 0.03165977820754051,
+    "probe/z_to_action_seq_h20/valid_ratio": 0.979296863079071,
+    "probe/z_to_action_seq_h20_shuffled/l2": 0.20211628079414368,
+    "probe/z_to_action_seq_h20_shuffled/valid_ratio": 0.979296863079071,
+    "probe/z_to_action_t/l2": 0.018910784274339676,
+    "probe/z_to_action_t_shuffled/l2": 0.19590125977993011,
+    "time/data_wait_s": 3.6241065219510347,
+    "time/iter_s": 7.669815575005487,
+    "time/step_s": 4.045709053054452
+  },
+  "best_steps": {
+    "loss/fdm_fit_loss": 17800,
+    "loss/fdm_h20": 17800,
+    "loss/total_loss": 17800,
+    "probe/z_to_action_seq_h20/l2": 19770,
+    "probe/z_to_action_seq_h20/valid_ratio": 2800,
+    "probe/z_to_action_seq_h20_shuffled/l2": 18950,
+    "probe/z_to_action_seq_h20_shuffled/valid_ratio": 2800,
+    "probe/z_to_action_t/l2": 18330,
+    "probe/z_to_action_t_shuffled/l2": 15545,
+    "time/data_wait_s": 0,
+    "time/iter_s": 0,
+    "time/step_s": 0
+  },
+  "flush_count": 20000,
+  "last_metrics": {
+    "loss/fdm_fit_loss": 0.27370309829711914,
+    "loss/fdm_h20": 0.27370309829711914,
+    "loss/total_loss": 0.27370309829711914,
+    "probe/z_to_action_seq_h20/l2": 0.03549542278051376,
+    "probe/z_to_action_seq_h20/valid_ratio": 0.944140613079071,
+    "probe/z_to_action_seq_h20_shuffled/l2": 0.21182116866111755,
+    "probe/z_to_action_seq_h20_shuffled/valid_ratio": 0.944140613079071,
+    "probe/z_to_action_t/l2": 0.027867570519447327,
+    "probe/z_to_action_t_shuffled/l2": 0.2077900469303131,
+    "time/data_wait_s": 0.0005364101380109787,
+    "time/iter_s": 0.4869440640322864,
+    "time/step_s": 0.4864076538942754
+  },
+  "last_step": 19999,
+  "paths": {
+    "artifacts_dir": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/stages/stage1/artifacts",
+    "metrics_jsonl": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/stages/stage1/metrics.jsonl",
+    "parent_run_json": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/stages/stage1/parent_run.json"
+  },
+  "stage": "stage1",
+  "stage_metadata": {
+    "checkpoint_path": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/stages/stage1",
+    "run_name": "latent_action_training_stage1_latent_action",
+    "status": "completed",
+    "tags": [
+      "debug"
+    ],
+    "wandb_project": "latent_action_stage1",
+    "wandb_run_id": "qudlwnz5",
+    "wandb_run_url": "https://wandb.ai/joonstack/latent_action_stage1/runs/qudlwnz5"
+  },
+  "status": "completed",
+  "updated_at": "2026-03-25T22:44:40.659757+00:00"
+}

wandb/debug.log

@@ -0,0 +1,24 @@
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_setup.py:_flush():81] Current SDK version is 0.25.1
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_setup.py:_flush():81] Configure stats pid to 3245764
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_init.py:setup_run_log_directory():717] Logging user logs to /home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/wandb/run-20260325_200057-qudlwnz5/logs/debug.log
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_init.py:setup_run_log_directory():718] Logging internal logs to /home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/wandb/run-20260325_200057-qudlwnz5/logs/debug-internal.log
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_init.py:init():844] calling init triggers
+2026-03-25 20:00:57,305 INFO    MainThread:3245764 [wandb_init.py:init():849] wandb.init called with sweep_config: {}
+config: {'data': {'name': 'libero_cosmos_policy', 'type': 'mg', 'backend': 'robomimic', 'paths': ['/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE3_turn_on_the_stove_and_put_the_moka_pot_on_it_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE4_put_the_black_bowl_in_the_bottom_drawer_of_the_cabinet_and_close_it_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE6_put_the_yellow_and_white_mug_in_the_microwave_and_close_it_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE8_put_both_moka_pots_on_the_stove_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE1_put_both_the_alphabet_soup_and_the_cream_cheese_box_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE2_put_both_the_alphabet_soup_and_the_tomato_sauce_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE2_put_both_the_cream_cheese_box_and_the_butter_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE5_put_the_white_mug_on_the_left_plate_and_put_the_yellow_and_white_mug_on_the_right_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE6_put_the_white_mug_on_the_plate_and_put_the_chocolate_pudding_to_the_right_of_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/STUDY_SCENE1_pick_up_the_book_and_place_it_in_the_back_compartment_of_the_caddy_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/open_the_middle_drawer_of_the_cabinet_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/open_the_top_drawer_and_put_the_bowl_inside_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/push_the_plate_to_the_front_of_the_stove_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_the_stove_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_top_of_the_cabinet_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_cream_cheese_in_the_bowl_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_wine_bottle_on_the_rack_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_wine_bottle_on_top_of_the_cabinet_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/turn_on_the_stove_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_alphabet_soup_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_bbq_sauce_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_butter_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_chocolate_pudding_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_cream_cheese_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_ketchup_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_milk_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_orange_juice_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_salad_dressing_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_tomato_sauce_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_between_the_plate_and_the_ramekin_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_from_table_center_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_in_the_top_drawer_of_the_wooden_cabinet_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_cookie_box_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_plate_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_ramekin_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_cookie_box_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_ramekin_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_stove_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_wooden_cabinet_and_place_it_on_the_plate_demo.hdf5'], 'task_suite_name': None, 'observations_keys': ['image'], 'observation_source_keys': {'image': 'agentview_rgb_jpeg'}, 'action_dim': 7, 'frame_stack': 1, 'horizon': 1, 'temporal_index_mode': 'clip', 'use_proprio': False, 'proprio_source': 'robot_states', 'image_chw': [3, 224, 224], 'image_value_range': 'zero_to_one', 'image_transport_dtype': 'uint8', 'duration_focus': None, 'hf': {'repo_id': 'nvidia/LIBERO-Cosmos-Policy', 'repo_type': 'dataset', 'allow_patterns': ['success_only/*_regen/*.hdf5'], 'local_files_only': False}, 'action_sequence_targets': {'enabled': True, 'horizons': [20], 'target_key': 'gt_action_seq_max'}, 'temporal': {'max_offset': 20, 'anchor_mode': 'fixed', 'extra_random_count': 0, 'load_full_context': False}, 'multiscale': {'enabled': True}, 'use_aug': False, 'emit_states': False, 'emit_duration_label': False, 'emit_future_offset_fields': False, 'sample_transform': {'weak_augmentation': {'enabled': False}}}, 'train': {'optimizer': {'type': 'adamw', 'lr': 0.0001, 'weight_decay': 1e-06}, 'batch_size': 256, 'num_workers': 4, 'prefetch_factor': None, 'log_interval': 100, 'total_steps': 5000, 'eval_interval': 500, 'use_aug': True, 'aug_mode': 'default', 'use_amp': True, 'device_transfer_non_blocking': True, 'seed': 42, 'scheduler': {'type': 'cosine', 'warmup_steps': 500, 'num_cycles': 0.5, 'min_lr_scale': 0.0}, 'decoder_dataloader': {'enabled': False, 'batch_size': 64, 'num_workers': 4, 'prefetch_factor': None, 'shuffle': True}, 'stages': {'stage1': True, 'stage2': False, 'stage3': False}, 'stage1': {'batch_size': 256, 'num_workers': 16, 'prefetch_factor': 2, 'log_interval': 100, 'total_steps': 20000, 'eval_interval': 500, 'optimizer': {'lr': 0.0001, 'weight_decay': 1e-06}}, 'stage2': {'batch_size': 256, 'num_workers': 4, 'prefetch_factor': None, 'log_interval': 100, 'total_steps': 5000, 'eval_interval': 500, 'latent_target': 'auto', 'optimizer': {'lr': 0.0001, 'weight_decay': 1e-06}, 'action_probe': {'enabled': True}}, 'stage3': {'batch_size': 256, 'num_workers': 4, 'prefetch_factor': None, 'log_interval': 100, 'total_steps': 5000, 'eval_interval': 500, 'final_eval_num_episodes': 100, 'log_eval_video': False, 'optimizer': {'lr': 0.0001, 'weight_decay': 1e-06}, 'trainable_scope': 'all', 'non_decoder_lr_scale': 1.0}, 'checkpoint': {'enabled': True, 'base_dir': None, 'save_interval': 1000, 'save_last': True, 'latest_only': True, 'load': {'stage1': {'mode': 'none', 'path': None}, 'stage2': {'mode': 'handoff', 'path': None, 'teacher_path': None}, 'stage3': {'mode': 'handoff', 'path': None}}}}, 'model': {'idm': {'type': 'token_idm', 'action_dim': 128, 'token_dim': 768, 'model_dim': 512, 'latent_dim': 32, 'num_action_tokens': 4, 'num_blocks': 4, 'num_heads': 8, 'dropout': 0.0}, 'fdm': {'type': 'token_fdm_duration', 'action_dim': 128, 'token_dim': 768, 'model_dim': 512, 'num_action_tokens': 4, 'num_blocks': 4, 'num_heads': 8, 'dropout': 0.0, 'prediction_mode': 'direct'}, 'hyperbolic_latent': {'enabled': False, 'backend': 'geoopt', 'manifold': 'poincare', 'curvature': 1.0, 'learnable_curvature': False, 'fdm_input_mode': 'logmap0', 'prelift_mode': 'none', 'prelift_scale': 1.0, 'lift_max_norm': 5.0, 'tangent_max_norm': 5.0, 'eps': 1e-06}, 'encoders': {'type': 'dino', 'image_key': 'image', 'model_id': 'facebook/dinov2-base', 'input_value_range': 'zero_to_one', 'freeze_backbone': True, 'drop_cls_token': True, 'output_dim': 768, 'mean': [0.485, 0.456, 0.406], 'std': [0.229, 0.224, 0.225]}, 'type': 'dino_lam', 'idm_input': 'future', 'fdm_target': 'future', 'pixel_decoders': None, 'latent_action_decoders': None, 'temporal': {'max_offset': 20, 'anchor_mode': 'fixed', 'extra_random_count': 0, 'load_full_context': False}, 'multiscale_loss': {'enabled': True, 'prediction_mode': 'direct_duration', 'weight_mode': 'uniform', 'weights': {}, 'allow_plain_fdm_oneshot': True, 'direct_duration_backend': 'duration_aware', 'consistency': {'enabled': False, 'weight': 10.0, 'num_pairs': 4, 'sample_source': 'all_horizons', 'teacher_mode': 'direct_teacher'}}, 'branch_order': {'enabled': False, 'radial_weight': 0.0, 'local_radial_margin_weight': 0.0, 'local_radial_margin_alpha': 0.05, 'branch_weight': 0.0, 'z0_origin_weight': 0.0, 'prefix_weight': 0.0, 'radius_progress_weight': 1.0, 'radius_progress_mode': 'offset_margin', 'radius_progress_alpha': 0.02, 'branch_margin_deg': 10.0, 'branch_margin_rad': 0.17453292519943295, 'eps': 1e-06}}, 'objective': {'fdm_target': 'future', 'idm_input': 'future', 'multiscale': {'enabled': True, 'consistency': {'enabled': False, 'weight': 10.0, 'num_pairs': 4, 'sample_source': 'all_horizons', 'teacher_mode': 'direct_teacher'}, 'prediction_mode': 'direct_duration', 'allow_plain_fdm_oneshot': True, 'weight_mode': 'uniform', 'weights': {}}, 'temporal': {'max_offset': 20, 'anchor_mode': 'fixed', 'extra_random_count': 0, 'load_full_context': False}, 'branch_order': {'enabled': False, 'radial_weight': 0.0, 'local_radial_margin_weight': 0.0, 'local_radial_margin_alpha': 0.05, 'branch_weight': 0.0, 'z0_origin_weight': 0.0, 'prefix_weight': 0.0, 'radius_progress_weight': 1.0, 'radius_progress_mode': 'offset_margin', 'radius_progress_alpha': 0.02, 'branch_margin_deg': 10.0, 'eps': 1e-06}, 'latent_plan': {'enabled': False, 'total_horizon': 20, 'num_chunks': 1, 'chunk_dim': 128, 'plan_dim': 128}}, 'eval': {'type': 'robomimic', 'data_path': [], 'reset_mode': 'env_reset', 'num_eval_episodes': 20, 'max_steps': 500, 'record_video': True, 'checkpoint_path': None, 'checkpoint_strict': True, 'use_checkpoint_cfg': True, 'video_output_path': None, 'obs_keys': None, 'use_proprio': None, 'use_object': False, 'resize_hw': None, 'image_value_range': None, 'seed': 0}, 'agent': {'encoders': {'image': {'in_channels': 3, 'output_dim': 512, 'output_mode': 'global', 'type': 'resnet18', 'pretrained': False}, 'proprio': {'input_dim': 9, 'hidden_dim': 128, 'output_dim': 64}, 'type': 'group', 'modalities': ['agentview_image', 'robot0_eye_in_hand_image'], 'proj_dim': 128}, 'policies': {'decoder': {'type': 'mlp', 'hidden_dims': [256, 256]}, 'hidden_dims': [512, 512, 256, 64], 'action_dim': 256, 'emb_dim': 384, 'gt_action_dim': 7}, 'type': 'latent_action'}, 'probes': {'enabled': True, 'every': 10, 'steps_per_call': 1, 'sequence': {'enabled': True, 'horizons': [20], 'target_key': 'gt_action_seq_max'}, 'list': {'z_to_s_t': {'name': 'z_to_s_t', 'enabled': False, 'shuffle': True, 'type': 'regression', 'input': 'z_t', 'target': 's_t', 'loss': 'mse', 'lr': 0.001, 'every': 10, 'mlp': {'hidden_dims': [128, 64]}}, 'z_to_s_tp': {'name': 'z_to_s_tp', 'enabled': False, 'shuffle': True, 'type': 'regression', 'input': 'z_t', 'target': 's_tp', 'loss': 'mse', 'lr': 0.001, 'every': 10, 'mlp': {'hidden_dims': [128, 64]}}, 'z_to_action_t': {'name': 'z_to_action_t', 'probe_type': 'z_to_action', 'enabled': True, 'shuffle': True, 'type': 'regression', 'input': 'z_t', 'target': 'gt_action', 'loss': 'mse', 'lr': 0.001, 'every': 1, 'mlp': {'hidden_dims': [128, 64]}}, 'z_to_action_seq_h20': {'name': 'z_to_action_seq_h20', 'probe_type': 'z_to_action_sequence', 'enabled': True, 'shuffle': True, 'type': 'regression', 'input': 'z_t', 'target': 'gt_action_seq_max', 'sequence_horizon': 20, 'loss': 'mse', 'lr': 0.001, 'every': 10, 'mlp': {'hidden_dims': [128, 64]}, 'output_dim': 140}}}, 'logger': {'project': 'latent_action', 'run_name': 'latent_action_training', 'tags': ['debug']}, 'output_root': 'outputs/token_fdm_duration/dino_suite_all_k20_extra0', '_wandb': {}}
+2026-03-25 20:00:57,305 INFO    MainThread:3245764 [wandb_init.py:init():892] starting backend
+2026-03-25 20:00:57,521 INFO    MainThread:3245764 [wandb_init.py:init():895] sending inform_init request
+2026-03-25 20:00:57,524 INFO    MainThread:3245764 [wandb_init.py:init():903] backend started and connected
+2026-03-25 20:00:57,526 INFO    MainThread:3245764 [wandb_init.py:init():973] updated telemetry
+2026-03-25 20:00:57,532 INFO    MainThread:3245764 [wandb_init.py:init():997] communicating run to backend with 90.0 second timeout
+2026-03-25 20:00:58,418 INFO    MainThread:3245764 [wandb_init.py:init():1042] starting run threads in backend
+2026-03-25 20:00:58,525 INFO    MainThread:3245764 [wandb_run.py:_console_start():2524] atexit reg
+2026-03-25 20:00:58,525 INFO    MainThread:3245764 [wandb_run.py:_redirect():2373] redirect: wrap_raw
+2026-03-25 20:00:58,525 INFO    MainThread:3245764 [wandb_run.py:_redirect():2442] Wrapping output streams.
+2026-03-25 20:00:58,525 INFO    MainThread:3245764 [wandb_run.py:_redirect():2465] Redirects installed.
+2026-03-25 20:00:58,527 INFO    MainThread:3245764 [wandb_init.py:init():1082] run started, returning control to user process
+2026-03-25 22:44:40,660 INFO    MainThread:3245764 [wandb_run.py:_finish():2291] finishing run joonstack/latent_action_stage1/qudlwnz5
+2026-03-25 22:44:40,661 INFO    MainThread:3245764 [wandb_run.py:_atexit_cleanup():2490] got exitcode: 0
+2026-03-25 22:44:40,662 INFO    MainThread:3245764 [wandb_run.py:_restore():2472] restore
+2026-03-25 22:44:40,662 INFO    MainThread:3245764 [wandb_run.py:_restore():2478] restore done
+2026-03-25 22:44:41,910 INFO    MainThread:3245764 [wandb_run.py:_footer_sync_info():3868] logging synced files

wandb/run-20260325_200057-qudlwnz5/files/config.yaml

@@ -0,0 +1,524 @@
+_wandb:
+    value:
+        cli_version: 0.25.1
+        e:
+            zgphb5mlnbcfb9jpehhh8zgxm0bu2zcl:
+                args:
+                    - --config-name
+                    - dino_lam
+                    - data=libero_cosmos_policy
+                    - data.task_suite_name=null
+                    - data.hf.allow_patterns=[success_only/*_regen/*.hdf5]
+                    - data.image_chw=[3,224,224]
+                    - data.temporal_index_mode=clip
+                    - model.fdm.type=token_fdm_duration
+                    - objective.idm_input=future
+                    - objective.temporal.max_offset=20
+                    - objective.temporal.anchor_mode=fixed
+                    - objective.temporal.extra_random_count=0
+                    - train.use_aug=true
+                    - train.use_amp=true
+                    - train.stage1.batch_size=256
+                    - train.stage1.num_workers=16
+                    - train.stage1.prefetch_factor=2
+                    - train.stage1.total_steps=20000
+                    - train.stage1.log_interval=100
+                    - train.checkpoint.enabled=true
+                    - train.checkpoint.save_interval=1000
+                    - train.checkpoint.save_last=true
+                    - train.checkpoint.latest_only=true
+                    - output_root=outputs/token_fdm_duration/dino_suite_all_k20_extra0
+                codePath: run.py
+                cpu_count: 32
+                cpu_count_logical: 64
+                cudaVersion: "12.4"
+                disk:
+                    /:
+                        total: "1889053200384"
+                        used: "1645352988672"
+                email: zuni1218@gmail.com
+                executable: /home/jovyan/youngjoonjeong/latent_action/.venv/bin/python3
+                git:
+                    commit: 84bdcca892435da456b8a676dae358493a0a6e59
+                    remote: https://github.com/joon-stack/latent_action.git
+                gpu: NVIDIA H100 NVL
+                gpu_count: 1
+                gpu_nvidia:
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "100485038080"
+                      name: NVIDIA H100 NVL
+                      uuid: GPU-c6a0d739-1597-848e-b639-5dba9b4deb79
+                host: 91b1e4a5838d
+                memory:
+                    total: "540748169216"
+                os: Linux-5.15.0-171-generic-x86_64-with-glibc2.35
+                program: /home/jovyan/youngjoonjeong/latent_action/run.py
+                python: CPython 3.10.18
+                root: /home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52
+                startedAt: "2026-03-25T20:00:57.302961Z"
+                writerId: zgphb5mlnbcfb9jpehhh8zgxm0bu2zcl
+        m: []
+        python_version: 3.10.18
+        t:
+            "1":
+                - 1
+                - 5
+                - 11
+                - 41
+                - 49
+                - 50
+                - 53
+                - 83
+            "2":
+                - 1
+                - 5
+                - 11
+                - 41
+                - 49
+                - 50
+                - 53
+                - 83
+            "3":
+                - 2
+                - 13
+                - 15
+                - 16
+                - 61
+            "4": 3.10.18
+            "5": 0.25.1
+            "6": 4.41.2
+            "10":
+                - 21
+            "12": 0.25.1
+            "13": linux-x86_64
+agent:
+    value:
+        encoders:
+            image:
+                in_channels: 3
+                output_dim: 512
+                output_mode: global
+                pretrained: false
+                type: resnet18
+            modalities:
+                - agentview_image
+                - robot0_eye_in_hand_image
+            proj_dim: 128
+            proprio:
+                hidden_dim: 128
+                input_dim: 9
+                output_dim: 64
+            type: group
+        policies:
+            action_dim: 256
+            decoder:
+                hidden_dims:
+                    - 256
+                    - 256
+                type: mlp
+            emb_dim: 384
+            gt_action_dim: 7
+            hidden_dims:
+                - 512
+                - 512
+                - 256
+                - 64
+        type: latent_action
+data:
+    value:
+        action_dim: 7
+        action_sequence_targets:
+            enabled: true
+            horizons:
+                - 20
+            target_key: gt_action_seq_max
+        backend: robomimic
+        duration_focus: null
+        emit_duration_label: false
+        emit_future_offset_fields: false
+        emit_states: false
+        frame_stack: 1
+        hf:
+            allow_patterns:
+                - success_only/*_regen/*.hdf5
+            local_files_only: false
+            repo_id: nvidia/LIBERO-Cosmos-Policy
+            repo_type: dataset
+        horizon: 1
+        image_chw:
+            - 3
+            - 224
+            - 224
+        image_transport_dtype: uint8
+        image_value_range: zero_to_one
+        multiscale:
+            enabled: true
+        name: libero_cosmos_policy
+        observation_source_keys:
+            image: agentview_rgb_jpeg
+        observations_keys:
+            - image
+        paths:
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE3_turn_on_the_stove_and_put_the_moka_pot_on_it_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE4_put_the_black_bowl_in_the_bottom_drawer_of_the_cabinet_and_close_it_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE6_put_the_yellow_and_white_mug_in_the_microwave_and_close_it_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE8_put_both_moka_pots_on_the_stove_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE1_put_both_the_alphabet_soup_and_the_cream_cheese_box_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE2_put_both_the_alphabet_soup_and_the_tomato_sauce_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE2_put_both_the_cream_cheese_box_and_the_butter_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE5_put_the_white_mug_on_the_left_plate_and_put_the_yellow_and_white_mug_on_the_right_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE6_put_the_white_mug_on_the_plate_and_put_the_chocolate_pudding_to_the_right_of_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/STUDY_SCENE1_pick_up_the_book_and_place_it_in_the_back_compartment_of_the_caddy_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/open_the_middle_drawer_of_the_cabinet_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/open_the_top_drawer_and_put_the_bowl_inside_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/push_the_plate_to_the_front_of_the_stove_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_the_stove_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_top_of_the_cabinet_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_cream_cheese_in_the_bowl_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_wine_bottle_on_the_rack_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_wine_bottle_on_top_of_the_cabinet_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/turn_on_the_stove_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_alphabet_soup_and_place_it_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_bbq_sauce_and_place_it_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_butter_and_place_it_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_chocolate_pudding_and_place_it_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_cream_cheese_and_place_it_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_ketchup_and_place_it_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_milk_and_place_it_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_orange_juice_and_place_it_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_salad_dressing_and_place_it_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_tomato_sauce_and_place_it_in_the_basket_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_between_the_plate_and_the_ramekin_and_place_it_on_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_from_table_center_and_place_it_on_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_in_the_top_drawer_of_the_wooden_cabinet_and_place_it_on_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_cookie_box_and_place_it_on_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_plate_and_place_it_on_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_ramekin_and_place_it_on_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_cookie_box_and_place_it_on_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_ramekin_and_place_it_on_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_stove_and_place_it_on_the_plate_demo.hdf5
+            - /home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_wooden_cabinet_and_place_it_on_the_plate_demo.hdf5
+        proprio_source: robot_states
+        sample_transform:
+            weak_augmentation:
+                enabled: false
+        task_suite_name: null
+        temporal:
+            anchor_mode: fixed
+            extra_random_count: 0
+            load_full_context: false
+            max_offset: 20
+        temporal_index_mode: clip
+        type: mg
+        use_aug: false
+        use_proprio: false
+eval:
+    value:
+        checkpoint_path: null
+        checkpoint_strict: true
+        data_path: []
+        image_value_range: null
+        max_steps: 500
+        num_eval_episodes: 20
+        obs_keys: null
+        record_video: true
+        reset_mode: env_reset
+        resize_hw: null
+        seed: 0
+        type: robomimic
+        use_checkpoint_cfg: true
+        use_object: false
+        use_proprio: null
+        video_output_path: null
+logger:
+    value:
+        project: latent_action
+        run_name: latent_action_training
+        tags:
+            - debug
+model:
+    value:
+        branch_order:
+            branch_margin_deg: 10
+            branch_margin_rad: 0.17453292519943295
+            branch_weight: 0
+            enabled: false
+            eps: 1e-06
+            local_radial_margin_alpha: 0.05
+            local_radial_margin_weight: 0
+            prefix_weight: 0
+            radial_weight: 0
+            radius_progress_alpha: 0.02
+            radius_progress_mode: offset_margin
+            radius_progress_weight: 1
+            z0_origin_weight: 0
+        encoders:
+            drop_cls_token: true
+            freeze_backbone: true
+            image_key: image
+            input_value_range: zero_to_one
+            mean:
+                - 0.485
+                - 0.456
+                - 0.406
+            model_id: facebook/dinov2-base
+            output_dim: 768
+            std:
+                - 0.229
+                - 0.224
+                - 0.225
+            type: dino
+        fdm:
+            action_dim: 128
+            dropout: 0
+            model_dim: 512
+            num_action_tokens: 4
+            num_blocks: 4
+            num_heads: 8
+            prediction_mode: direct
+            token_dim: 768
+            type: token_fdm_duration
+        fdm_target: future
+        hyperbolic_latent:
+            backend: geoopt
+            curvature: 1
+            enabled: false
+            eps: 1e-06
+            fdm_input_mode: logmap0
+            learnable_curvature: false
+            lift_max_norm: 5
+            manifold: poincare
+            prelift_mode: none
+            prelift_scale: 1
+            tangent_max_norm: 5
+        idm:
+            action_dim: 128
+            dropout: 0
+            latent_dim: 32
+            model_dim: 512
+            num_action_tokens: 4
+            num_blocks: 4
+            num_heads: 8
+            token_dim: 768
+            type: token_idm
+        idm_input: future
+        latent_action_decoders: null
+        multiscale_loss:
+            allow_plain_fdm_oneshot: true
+            consistency:
+                enabled: false
+                num_pairs: 4
+                sample_source: all_horizons
+                teacher_mode: direct_teacher
+                weight: 10
+            direct_duration_backend: duration_aware
+            enabled: true
+            prediction_mode: direct_duration
+            weight_mode: uniform
+        pixel_decoders: null
+        temporal:
+            anchor_mode: fixed
+            extra_random_count: 0
+            load_full_context: false
+            max_offset: 20
+        type: dino_lam
+objective:
+    value:
+        branch_order:
+            branch_margin_deg: 10
+            branch_weight: 0
+            enabled: false
+            eps: 1e-06
+            local_radial_margin_alpha: 0.05
+            local_radial_margin_weight: 0
+            prefix_weight: 0
+            radial_weight: 0
+            radius_progress_alpha: 0.02
+            radius_progress_mode: offset_margin
+            radius_progress_weight: 1
+            z0_origin_weight: 0
+        fdm_target: future
+        idm_input: future
+        latent_plan:
+            chunk_dim: 128
+            enabled: false
+            num_chunks: 1
+            plan_dim: 128
+            total_horizon: 20
+        multiscale:
+            allow_plain_fdm_oneshot: true
+            consistency:
+                enabled: false
+                num_pairs: 4
+                sample_source: all_horizons
+                teacher_mode: direct_teacher
+                weight: 10
+            enabled: true
+            prediction_mode: direct_duration
+            weight_mode: uniform
+        temporal:
+            anchor_mode: fixed
+            extra_random_count: 0
+            load_full_context: false
+            max_offset: 20
+output_root:
+    value: outputs/token_fdm_duration/dino_suite_all_k20_extra0
+probes:
+    value:
+        enabled: true
+        every: 10
+        list:
+            z_to_action_seq_h20:
+                enabled: true
+                every: 10
+                input: z_t
+                loss: mse
+                lr: 0.001
+                mlp:
+                    hidden_dims:
+                        - 128
+                        - 64
+                name: z_to_action_seq_h20
+                output_dim: 140
+                probe_type: z_to_action_sequence
+                sequence_horizon: 20
+                shuffle: true
+                target: gt_action_seq_max
+                type: regression
+            z_to_action_t:
+                enabled: true
+                every: 1
+                input: z_t
+                loss: mse
+                lr: 0.001
+                mlp:
+                    hidden_dims:
+                        - 128
+                        - 64
+                name: z_to_action_t
+                probe_type: z_to_action
+                shuffle: true
+                target: gt_action
+                type: regression
+            z_to_s_t:
+                enabled: false
+                every: 10
+                input: z_t
+                loss: mse
+                lr: 0.001
+                mlp:
+                    hidden_dims:
+                        - 128
+                        - 64
+                name: z_to_s_t
+                shuffle: true
+                target: s_t
+                type: regression
+            z_to_s_tp:
+                enabled: false
+                every: 10
+                input: z_t
+                loss: mse
+                lr: 0.001
+                mlp:
+                    hidden_dims:
+                        - 128
+                        - 64
+                name: z_to_s_tp
+                shuffle: true
+                target: s_tp
+                type: regression
+        sequence:
+            enabled: true
+            horizons:
+                - 20
+            target_key: gt_action_seq_max
+        steps_per_call: 1
+train:
+    value:
+        aug_mode: default
+        batch_size: 256
+        checkpoint:
+            base_dir: null
+            enabled: true
+            latest_only: true
+            load:
+                stage1:
+                    mode: none
+                    path: null
+                stage2:
+                    mode: handoff
+                    path: null
+                    teacher_path: null
+                stage3:
+                    mode: handoff
+                    path: null
+            save_interval: 1000
+            save_last: true
+        decoder_dataloader:
+            batch_size: 64
+            enabled: false
+            num_workers: 4
+            prefetch_factor: null
+            shuffle: true
+        device_transfer_non_blocking: true
+        eval_interval: 500
+        log_interval: 100
+        num_workers: 4
+        optimizer:
+            lr: 0.0001
+            type: adamw
+            weight_decay: 1e-06
+        prefetch_factor: null
+        scheduler:
+            min_lr_scale: 0
+            num_cycles: 0.5
+            type: cosine
+            warmup_steps: 500
+        seed: 42
+        stage1:
+            batch_size: 256
+            eval_interval: 500
+            log_interval: 100
+            num_workers: 16
+            optimizer:
+                lr: 0.0001
+                weight_decay: 1e-06
+            prefetch_factor: 2
+            total_steps: 20000
+        stage2:
+            action_probe:
+                enabled: true
+            batch_size: 256
+            eval_interval: 500
+            latent_target: auto
+            log_interval: 100
+            num_workers: 4
+            optimizer:
+                lr: 0.0001
+                weight_decay: 1e-06
+            prefetch_factor: null
+            total_steps: 5000
+        stage3:
+            batch_size: 256
+            eval_interval: 500
+            final_eval_num_episodes: 100
+            log_eval_video: false
+            log_interval: 100
+            non_decoder_lr_scale: 1
+            num_workers: 4
+            optimizer:
+                lr: 0.0001
+                weight_decay: 1e-06
+            prefetch_factor: null
+            total_steps: 5000
+            trainable_scope: all
+        stages:
+            stage1: true
+            stage2: false
+            stage3: false
+        total_steps: 5000
+        use_amp: true
+        use_aug: true

wandb/run-20260325_200057-qudlwnz5/files/output.log

@@ -0,0 +1,7 @@
+wandb: Detected [huggingface_hub.inference] in use.
+wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
+wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
+wandb: WARNING The get_url method is deprecated and will be removed in a future release. Please use `run.url` instead.
+[checkpoint] stage=stage1, path=/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/stages/stage1
+[checkpoint] stage=stage1, config_yaml=/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/stages/stage1/hydra_resolved.yaml
+                                                                                                                                                                        

wandb/run-20260325_200057-qudlwnz5/files/requirements.txt

@@ -0,0 +1,175 @@
+joblib==1.5.3
+fastjsonschema==2.21.2
+exceptiongroup==1.3.1
+fonttools==4.62.1
+easydict==1.13
+egl_probe==1.0.2
+xyzservices==2025.11.0
+Werkzeug==3.1.6
+decorator==5.2.1
+pytz==2026.1.post1
+regex==2026.2.28
+rpds-py==0.30.0
+scikit-learn==1.7.2
+tzdata==2025.3
+antlr4-python3-runtime==4.9.3
+evdev==1.9.3
+gitdb==4.0.12
+PyYAML==6.0.3
+tomli==2.4.0
+pandas==2.3.3
+soundfile==0.13.1
+tokenizers==0.19.1
+future==1.0.0
+attrs==26.1.0
+MarkupSafe==3.0.3
+platformdirs==4.9.4
+triton==3.1.0
+annotated-types==0.7.0
+rdkit==2025.9.6
+narwhals==2.18.0
+Markdown==3.10.2
+python-xlib==0.33
+transformers==4.41.2
+typing-inspection==0.4.2
+threadpoolctl==3.6.0
+traitlets==5.14.3
+pycparser==3.0
+python-dateutil==2.9.0.post0
+sentry-sdk==2.55.0
+nbformat==5.10.4
+aiohappyeyeballs==2.6.1
+GitPython==3.1.46
+pydantic==2.12.5
+pyarrow-hotfix==0.7
+ImageIO==2.37.3
+nvidia-cublas-cu12==12.4.5.8
+typing_extensions==4.15.0
+nvidia-cuda-cupti-cu12==12.4.127
+tensorboard-data-server==0.7.2
+importlib_resources==6.5.2
+cloudpickle==3.1.2
+pynput==1.8.1
+multidict==6.7.1
+diffusers==0.11.1
+iniconfig==2.3.0
+requests==2.32.5
+vector-quantize-pytorch==1.28.0
+nvidia-cudnn-cu12==9.1.0.70
+torchinfo==1.8.0
+matplotlib==3.9.4
+six==1.17.0
+tqdm==4.67.3
+nvidia-cuda-runtime-cu12==12.4.127
+wandb==0.25.1
+robomimic==0.5.0
+tensorboardX==2.6.4
+async-timeout==5.0.1
+python-dotenv==1.2.2
+mink==1.1.0
+nvidia-curand-cu12==10.3.5.147
+urllib3==2.6.3
+cffi==2.0.0
+nvidia-cusolver-cu12==11.6.1.9
+tensorboard==2.20.0
+nvidia-nvjitlink-cu12==12.4.127
+smmap==5.0.3
+datasets==2.19.2
+frozenlist==1.8.0
+nvidia-cusparse-cu12==12.3.1.170
+importlib_metadata==9.0.0
+gym-notices==0.1.0
+termcolor==3.3.0
+jupyter_core==5.9.1
+packaging==26.0
+PyOpenGL==3.1.10
+numpy==1.26.4
+jsonschema==4.26.0
+bokeh==3.9.0
+nvidia-nccl-cu12==2.21.5
+filelock==3.25.2
+torch==2.5.1
+nltk==3.9.3
+pytest==9.0.2
+aiosignal==1.4.0
+scipy==1.15.3
+daqp==0.8.3
+tornado==6.5.5
+proglog==0.1.12
+huggingface-hub==0.23.4
+pyparsing==3.3.2
+imageio-ffmpeg==0.6.0
+referencing==0.37.0
+pluggy==1.6.0
+einops==0.8.2
+yarl==1.23.0
+dill==0.3.8
+torchvision==0.20.1
+jsonschema-specifications==2025.9.1
+pip==23.0.1
+moviepy==2.2.1
+mdit-py-plugins==0.5.0
+etils==1.13.0
+nvidia-nvtx-cu12==12.4.127
+numba==0.64.0
+safetensors==0.7.0
+kiwisolver==1.5.0
+contourpy==1.3.2
+networkx==3.4.2
+cycler==0.12.1
+aiohttp==3.13.3
+zipp==3.23.0
+mpmath==1.3.0
+gym==0.26.2
+opencv-python==4.11.0.86
+pillow==11.3.0
+xxhash==3.6.0
+psutil==7.2.2
+hydra-core==1.3.2
+absl-py==2.4.0
+glfw==2.10.0
+idna==3.11
+einx==0.4.2
+Pygments==2.19.2
+PyOpenGL-accelerate==3.1.10
+frozendict==2.4.7
+sympy==1.13.1
+h5py==3.16.0
+Cython==0.29.37
+qpsolvers==4.11.0
+certifi==2026.2.25
+geoopt==0.5.1
+nvidia-cufft-cu12==11.2.1.3
+charset-normalizer==3.4.6
+mdurl==0.1.2
+pydantic_core==2.41.5
+nvidia-cuda-nvrtc-cu12==12.4.127
+protobuf==6.33.6
+grpcio==1.78.0
+jupytext==1.19.1
+robosuite==1.5.1
+bddl==3.6.0
+propcache==0.4.1
+Jinja2==3.1.6
+click==8.3.1
+markdown-it-py==4.0.0
+fsspec==2024.3.1
+omegaconf==2.3.0
+llvmlite==0.46.0
+setuptools==82.0.1
+pyarrow==19.0.1
+multiprocess==0.70.16
+mujoco==3.6.0
+plotly==6.6.0
+jaraco.functools==4.4.0
+platformdirs==4.4.0
+tomli==2.4.0
+backports.tarfile==1.2.0
+jaraco.context==6.1.0
+autocommand==2.2.2
+jaraco.text==4.0.0
+importlib_metadata==8.7.1
+more-itertools==10.8.0
+packaging==26.0
+zipp==3.23.0
+wheel==0.46.3

wandb/run-20260325_200057-qudlwnz5/files/wandb-metadata.json

@@ -0,0 +1,65 @@
+{
+  "os": "Linux-5.15.0-171-generic-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.18",
+  "startedAt": "2026-03-25T20:00:57.302961Z",
+  "args": [
+    "--config-name",
+    "dino_lam",
+    "data=libero_cosmos_policy",
+    "data.task_suite_name=null",
+    "data.hf.allow_patterns=[success_only/*_regen/*.hdf5]",
+    "data.image_chw=[3,224,224]",
+    "data.temporal_index_mode=clip",
+    "model.fdm.type=token_fdm_duration",
+    "objective.idm_input=future",
+    "objective.temporal.max_offset=20",
+    "objective.temporal.anchor_mode=fixed",
+    "objective.temporal.extra_random_count=0",
+    "train.use_aug=true",
+    "train.use_amp=true",
+    "train.stage1.batch_size=256",
+    "train.stage1.num_workers=16",
+    "train.stage1.prefetch_factor=2",
+    "train.stage1.total_steps=20000",
+    "train.stage1.log_interval=100",
+    "train.checkpoint.enabled=true",
+    "train.checkpoint.save_interval=1000",
+    "train.checkpoint.save_last=true",
+    "train.checkpoint.latest_only=true",
+    "output_root=outputs/token_fdm_duration/dino_suite_all_k20_extra0"
+  ],
+  "program": "/home/jovyan/youngjoonjeong/latent_action/run.py",
+  "codePath": "run.py",
+  "git": {
+    "remote": "https://github.com/joon-stack/latent_action.git",
+    "commit": "84bdcca892435da456b8a676dae358493a0a6e59"
+  },
+  "email": "zuni1218@gmail.com",
+  "root": "/home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52",
+  "host": "91b1e4a5838d",
+  "executable": "/home/jovyan/youngjoonjeong/latent_action/.venv/bin/python3",
+  "cpu_count": 32,
+  "cpu_count_logical": 64,
+  "gpu": "NVIDIA H100 NVL",
+  "gpu_count": 1,
+  "disk": {
+    "/": {
+      "total": "1889053200384",
+      "used": "1645352988672"
+    }
+  },
+  "memory": {
+    "total": "540748169216"
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper",
+      "uuid": "GPU-c6a0d739-1597-848e-b639-5dba9b4deb79"
+    }
+  ],
+  "cudaVersion": "12.4",
+  "writerId": "zgphb5mlnbcfb9jpehhh8zgxm0bu2zcl"
+}

wandb/run-20260325_200057-qudlwnz5/files/wandb-summary.json

@@ -0,0 +1 @@
+{"time/data_wait_s":0.0005364101380109787,"loss/total_loss":0.27370309829711914,"probe/z_to_action_seq_h20/l2":0.03549542278051376,"_timestamp":1.77447867603798e+09,"time/iter_s":0.4869440640322864,"_wandb":{"runtime":9822},"probe/z_to_action_seq_h20_shuffled/valid_ratio":0.944140613079071,"time/step_s":0.4864076538942754,"probe/z_to_action_t_shuffled/l2":0.2077900469303131,"probe/z_to_action_seq_h20_shuffled/l2":0.21182116866111755,"probe/z_to_action_t/l2":0.027867570519447327,"loss/fdm_fit_loss":0.27370309829711914,"loss/fdm_h20":0.27370309829711914,"probe/z_to_action_seq_h20/valid_ratio":0.944140613079071,"_step":19999,"_runtime":9822.244018516}

wandb/run-20260325_200057-qudlwnz5/logs/debug-core.log

@@ -0,0 +1,19 @@
+{"time":"2026-03-25T20:00:57.333745886Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpr17fz98x/port-3245764.txt","pid":3245764,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-03-25T20:00:57.334247558Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":3245764}
+{"time":"2026-03-25T20:00:57.334229645Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-3245764-3246101-3854211172/socket","Net":"unix"}}
+{"time":"2026-03-25T20:00:57.521004148Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-03-25T20:00:57.525759429Z","level":"INFO","msg":"handleInformInit: received","streamId":"qudlwnz5","id":"1(@)"}
+{"time":"2026-03-25T20:00:57.863985537Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"qudlwnz5","id":"1(@)"}
+{"time":"2026-03-25T20:01:03.533019493Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"2webd6hskw9a"}
+{"time":"2026-03-25T22:44:40.662595918Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"2webd6hskw9a"}
+{"time":"2026-03-25T22:44:41.91056233Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"2webd6hskw9a"}
+{"time":"2026-03-25T22:44:41.91105048Z","level":"INFO","msg":"handleInformFinish: finish message received","streamId":"qudlwnz5","id":"1(@)"}
+{"time":"2026-03-25T22:44:41.912031578Z","level":"INFO","msg":"handleInformFinish: stream closed","streamId":"qudlwnz5","id":"1(@)"}
+{"time":"2026-03-25T22:44:42.151131844Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2026-03-25T22:44:42.151173852Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2026-03-25T22:44:42.151183039Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2026-03-25T22:44:42.151208807Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2026-03-25T22:44:42.151268228Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-3245764-3246101-3854211172/socket","Net":"unix"}}
+{"time":"2026-03-25T22:44:42.151299827Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2026-03-25T22:44:42.151310276Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2026-03-25T22:44:42.151320816Z","level":"INFO","msg":"server is closed"}

wandb/run-20260325_200057-qudlwnz5/logs/debug.log

@@ -0,0 +1,24 @@
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_setup.py:_flush():81] Current SDK version is 0.25.1
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_setup.py:_flush():81] Configure stats pid to 3245764
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_init.py:setup_run_log_directory():717] Logging user logs to /home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/wandb/run-20260325_200057-qudlwnz5/logs/debug.log
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_init.py:setup_run_log_directory():718] Logging internal logs to /home/jovyan/youngjoonjeong/latent_action/outputs/token_fdm_duration/dino_suite_all_k20_extra0/libero_cosmos_policy/2026-03-25/20-00-52/wandb/run-20260325_200057-qudlwnz5/logs/debug-internal.log
+2026-03-25 20:00:57,304 INFO    MainThread:3245764 [wandb_init.py:init():844] calling init triggers
+2026-03-25 20:00:57,305 INFO    MainThread:3245764 [wandb_init.py:init():849] wandb.init called with sweep_config: {}
+config: {'data': {'name': 'libero_cosmos_policy', 'type': 'mg', 'backend': 'robomimic', 'paths': ['/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE3_turn_on_the_stove_and_put_the_moka_pot_on_it_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE4_put_the_black_bowl_in_the_bottom_drawer_of_the_cabinet_and_close_it_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE6_put_the_yellow_and_white_mug_in_the_microwave_and_close_it_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/KITCHEN_SCENE8_put_both_moka_pots_on_the_stove_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE1_put_both_the_alphabet_soup_and_the_cream_cheese_box_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE2_put_both_the_alphabet_soup_and_the_tomato_sauce_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE2_put_both_the_cream_cheese_box_and_the_butter_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE5_put_the_white_mug_on_the_left_plate_and_put_the_yellow_and_white_mug_on_the_right_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/LIVING_ROOM_SCENE6_put_the_white_mug_on_the_plate_and_put_the_chocolate_pudding_to_the_right_of_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_10_regen/STUDY_SCENE1_pick_up_the_book_and_place_it_in_the_back_compartment_of_the_caddy_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/open_the_middle_drawer_of_the_cabinet_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/open_the_top_drawer_and_put_the_bowl_inside_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/push_the_plate_to_the_front_of_the_stove_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_the_stove_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_bowl_on_top_of_the_cabinet_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_cream_cheese_in_the_bowl_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_wine_bottle_on_the_rack_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/put_the_wine_bottle_on_top_of_the_cabinet_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_goal_regen/turn_on_the_stove_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_alphabet_soup_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_bbq_sauce_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_butter_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_chocolate_pudding_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_cream_cheese_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_ketchup_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_milk_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_orange_juice_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_salad_dressing_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_object_regen/pick_up_the_tomato_sauce_and_place_it_in_the_basket_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_between_the_plate_and_the_ramekin_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_from_table_center_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_in_the_top_drawer_of_the_wooden_cabinet_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_cookie_box_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_plate_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_next_to_the_ramekin_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_cookie_box_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_ramekin_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_stove_and_place_it_on_the_plate_demo.hdf5', '/home/jovyan/.cache/huggingface/hub/datasets--nvidia--LIBERO-Cosmos-Policy/snapshots/b463cbd5d114909e4daea65eb03bed9787290042/success_only/libero_spatial_regen/pick_up_the_black_bowl_on_the_wooden_cabinet_and_place_it_on_the_plate_demo.hdf5'], 'task_suite_name': None, 'observations_keys': ['image'], 'observation_source_keys': {'image': 'agentview_rgb_jpeg'}, 'action_dim': 7, 'frame_stack': 1, 'horizon': 1, 'temporal_index_mode': 'clip', 'use_proprio': False, 'proprio_source': 'robot_states', 'image_chw': [3, 224, 224], 'image_value_range': 'zero_to_one', 'image_transport_dtype': 'uint8', 'duration_focus': None, 'hf': {'repo_id': 'nvidia/LIBERO-Cosmos-Policy', 'repo_type': 'dataset', 'allow_patterns': ['success_only/*_regen/*.hdf5'], 'local_files_only': False}, 'action_sequence_targets': {'enabled': True, 'horizons': [20], 'target_key': 'gt_action_seq_max'}, 'temporal': {'max_offset': 20, 'anchor_mode': 'fixed', 'extra_random_count': 0, 'load_full_context': False}, 'multiscale': {'enabled': True}, 'use_aug': False, 'emit_states': False, 'emit_duration_label': False, 'emit_future_offset_fields': False, 'sample_transform': {'weak_augmentation': {'enabled': False}}}, 'train': {'optimizer': {'type': 'adamw', 'lr': 0.0001, 'weight_decay': 1e-06}, 'batch_size': 256, 'num_workers': 4, 'prefetch_factor': None, 'log_interval': 100, 'total_steps': 5000, 'eval_interval': 500, 'use_aug': True, 'aug_mode': 'default', 'use_amp': True, 'device_transfer_non_blocking': True, 'seed': 42, 'scheduler': {'type': 'cosine', 'warmup_steps': 500, 'num_cycles': 0.5, 'min_lr_scale': 0.0}, 'decoder_dataloader': {'enabled': False, 'batch_size': 64, 'num_workers': 4, 'prefetch_factor': None, 'shuffle': True}, 'stages': {'stage1': True, 'stage2': False, 'stage3': False}, 'stage1': {'batch_size': 256, 'num_workers': 16, 'prefetch_factor': 2, 'log_interval': 100, 'total_steps': 20000, 'eval_interval': 500, 'optimizer': {'lr': 0.0001, 'weight_decay': 1e-06}}, 'stage2': {'batch_size': 256, 'num_workers': 4, 'prefetch_factor': None, 'log_interval': 100, 'total_steps': 5000, 'eval_interval': 500, 'latent_target': 'auto', 'optimizer': {'lr': 0.0001, 'weight_decay': 1e-06}, 'action_probe': {'enabled': True}}, 'stage3': {'batch_size': 256, 'num_workers': 4, 'prefetch_factor': None, 'log_interval': 100, 'total_steps': 5000, 'eval_interval': 500, 'final_eval_num_episodes': 100, 'log_eval_video': False, 'optimizer': {'lr': 0.0001, 'weight_decay': 1e-06}, 'trainable_scope': 'all', 'non_decoder_lr_scale': 1.0}, 'checkpoint': {'enabled': True, 'base_dir': None, 'save_interval': 1000, 'save_last': True, 'latest_only': True, 'load': {'stage1': {'mode': 'none', 'path': None}, 'stage2': {'mode': 'handoff', 'path': None, 'teacher_path': None}, 'stage3': {'mode': 'handoff', 'path': None}}}}, 'model': {'idm': {'type': 'token_idm', 'action_dim': 128, 'token_dim': 768, 'model_dim': 512, 'latent_dim': 32, 'num_action_tokens': 4, 'num_blocks': 4, 'num_heads': 8, 'dropout': 0.0}, 'fdm': {'type': 'token_fdm_duration', 'action_dim': 128, 'token_dim': 768, 'model_dim': 512, 'num_action_tokens': 4, 'num_blocks': 4, 'num_heads': 8, 'dropout': 0.0, 'prediction_mode': 'direct'}, 'hyperbolic_latent': {'enabled': False, 'backend': 'geoopt', 'manifold': 'poincare', 'curvature': 1.0, 'learnable_curvature': False, 'fdm_input_mode': 'logmap0', 'prelift_mode': 'none', 'prelift_scale': 1.0, 'lift_max_norm': 5.0, 'tangent_max_norm': 5.0, 'eps': 1e-06}, 'encoders': {'type': 'dino', 'image_key': 'image', 'model_id': 'facebook/dinov2-base', 'input_value_range': 'zero_to_one', 'freeze_backbone': True, 'drop_cls_token': True, 'output_dim': 768, 'mean': [0.485, 0.456, 0.406], 'std': [0.229, 0.224, 0.225]}, 'type': 'dino_lam', 'idm_input': 'future', 'fdm_target': 'future', 'pixel_decoders': None, 'latent_action_decoders': None, 'temporal': {'max_offset': 20, 'anchor_mode': 'fixed', 'extra_random_count': 0, 'load_full_context': False}, 'multiscale_loss': {'enabled': True, 'prediction_mode': 'direct_duration', 'weight_mode': 'uniform', 'weights': {}, 'allow_plain_fdm_oneshot': True, 'direct_duration_backend': 'duration_aware', 'consistency': {'enabled': False, 'weight': 10.0, 'num_pairs': 4, 'sample_source': 'all_horizons', 'teacher_mode': 'direct_teacher'}}, 'branch_order': {'enabled': False, 'radial_weight': 0.0, 'local_radial_margin_weight': 0.0, 'local_radial_margin_alpha': 0.05, 'branch_weight': 0.0, 'z0_origin_weight': 0.0, 'prefix_weight': 0.0, 'radius_progress_weight': 1.0, 'radius_progress_mode': 'offset_margin', 'radius_progress_alpha': 0.02, 'branch_margin_deg': 10.0, 'branch_margin_rad': 0.17453292519943295, 'eps': 1e-06}}, 'objective': {'fdm_target': 'future', 'idm_input': 'future', 'multiscale': {'enabled': True, 'consistency': {'enabled': False, 'weight': 10.0, 'num_pairs': 4, 'sample_source': 'all_horizons', 'teacher_mode': 'direct_teacher'}, 'prediction_mode': 'direct_duration', 'allow_plain_fdm_oneshot': True, 'weight_mode': 'uniform', 'weights': {}}, 'temporal': {'max_offset': 20, 'anchor_mode': 'fixed', 'extra_random_count': 0, 'load_full_context': False}, 'branch_order': {'enabled': False, 'radial_weight': 0.0, 'local_radial_margin_weight': 0.0, 'local_radial_margin_alpha': 0.05, 'branch_weight': 0.0, 'z0_origin_weight': 0.0, 'prefix_weight': 0.0, 'radius_progress_weight': 1.0, 'radius_progress_mode': 'offset_margin', 'radius_progress_alpha': 0.02, 'branch_margin_deg': 10.0, 'eps': 1e-06}, 'latent_plan': {'enabled': False, 'total_horizon': 20, 'num_chunks': 1, 'chunk_dim': 128, 'plan_dim': 128}}, 'eval': {'type': 'robomimic', 'data_path': [], 'reset_mode': 'env_reset', 'num_eval_episodes': 20, 'max_steps': 500, 'record_video': True, 'checkpoint_path': None, 'checkpoint_strict': True, 'use_checkpoint_cfg': True, 'video_output_path': None, 'obs_keys': None, 'use_proprio': None, 'use_object': False, 'resize_hw': None, 'image_value_range': None, 'seed': 0}, 'agent': {'encoders': {'image': {'in_channels': 3, 'output_dim': 512, 'output_mode': 'global', 'type': 'resnet18', 'pretrained': False}, 'proprio': {'input_dim': 9, 'hidden_dim': 128, 'output_dim': 64}, 'type': 'group', 'modalities': ['agentview_image', 'robot0_eye_in_hand_image'], 'proj_dim': 128}, 'policies': {'decoder': {'type': 'mlp', 'hidden_dims': [256, 256]}, 'hidden_dims': [512, 512, 256, 64], 'action_dim': 256, 'emb_dim': 384, 'gt_action_dim': 7}, 'type': 'latent_action'}, 'probes': {'enabled': True, 'every': 10, 'steps_per_call': 1, 'sequence': {'enabled': True, 'horizons': [20], 'target_key': 'gt_action_seq_max'}, 'list': {'z_to_s_t': {'name': 'z_to_s_t', 'enabled': False, 'shuffle': True, 'type': 'regression', 'input': 'z_t', 'target': 's_t', 'loss': 'mse', 'lr': 0.001, 'every': 10, 'mlp': {'hidden_dims': [128, 64]}}, 'z_to_s_tp': {'name': 'z_to_s_tp', 'enabled': False, 'shuffle': True, 'type': 'regression', 'input': 'z_t', 'target': 's_tp', 'loss': 'mse', 'lr': 0.001, 'every': 10, 'mlp': {'hidden_dims': [128, 64]}}, 'z_to_action_t': {'name': 'z_to_action_t', 'probe_type': 'z_to_action', 'enabled': True, 'shuffle': True, 'type': 'regression', 'input': 'z_t', 'target': 'gt_action', 'loss': 'mse', 'lr': 0.001, 'every': 1, 'mlp': {'hidden_dims': [128, 64]}}, 'z_to_action_seq_h20': {'name': 'z_to_action_seq_h20', 'probe_type': 'z_to_action_sequence', 'enabled': True, 'shuffle': True, 'type': 'regression', 'input': 'z_t', 'target': 'gt_action_seq_max', 'sequence_horizon': 20, 'loss': 'mse', 'lr': 0.001, 'every': 10, 'mlp': {'hidden_dims': [128, 64]}, 'output_dim': 140}}}, 'logger': {'project': 'latent_action', 'run_name': 'latent_action_training', 'tags': ['debug']}, 'output_root': 'outputs/token_fdm_duration/dino_suite_all_k20_extra0', '_wandb': {}}
+2026-03-25 20:00:57,305 INFO    MainThread:3245764 [wandb_init.py:init():892] starting backend
+2026-03-25 20:00:57,521 INFO    MainThread:3245764 [wandb_init.py:init():895] sending inform_init request
+2026-03-25 20:00:57,524 INFO    MainThread:3245764 [wandb_init.py:init():903] backend started and connected
+2026-03-25 20:00:57,526 INFO    MainThread:3245764 [wandb_init.py:init():973] updated telemetry
+2026-03-25 20:00:57,532 INFO    MainThread:3245764 [wandb_init.py:init():997] communicating run to backend with 90.0 second timeout
+2026-03-25 20:00:58,418 INFO    MainThread:3245764 [wandb_init.py:init():1042] starting run threads in backend
+2026-03-25 20:00:58,525 INFO    MainThread:3245764 [wandb_run.py:_console_start():2524] atexit reg
+2026-03-25 20:00:58,525 INFO    MainThread:3245764 [wandb_run.py:_redirect():2373] redirect: wrap_raw
+2026-03-25 20:00:58,525 INFO    MainThread:3245764 [wandb_run.py:_redirect():2442] Wrapping output streams.
+2026-03-25 20:00:58,525 INFO    MainThread:3245764 [wandb_run.py:_redirect():2465] Redirects installed.
+2026-03-25 20:00:58,527 INFO    MainThread:3245764 [wandb_init.py:init():1082] run started, returning control to user process
+2026-03-25 22:44:40,660 INFO    MainThread:3245764 [wandb_run.py:_finish():2291] finishing run joonstack/latent_action_stage1/qudlwnz5
+2026-03-25 22:44:40,661 INFO    MainThread:3245764 [wandb_run.py:_atexit_cleanup():2490] got exitcode: 0
+2026-03-25 22:44:40,662 INFO    MainThread:3245764 [wandb_run.py:_restore():2472] restore
+2026-03-25 22:44:40,662 INFO    MainThread:3245764 [wandb_run.py:_restore():2478] restore done
+2026-03-25 22:44:41,910 INFO    MainThread:3245764 [wandb_run.py:_footer_sync_info():3868] logging synced files

wandb/run-20260325_200057-qudlwnz5/run-qudlwnz5.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6838dfeb86abfb1f6ca625eb16a905847270272d4cd7b40c5c7758190e2037f
+size 22197604