Add files using upload-large-folder tool

.hydra/config.yaml

@@ -0,0 +1,205 @@
+model:
+  encoder:
+    backbone:
+      name: croco
+      model: ViTLarge_BaseDecoder
+      patch_embed_cls: PatchEmbedDust3R
+      asymmetry_decoder: true
+      intrinsics_embed_loc: encoder
+      intrinsics_embed_degree: 4
+      intrinsics_embed_type: token
+    name: noposplat
+    opacity_mapping:
+      initial: 0.0
+      final: 0.0
+      warm_up: 1
+    num_monocular_samples: 32
+    num_surfaces: 1
+    predict_opacity: false
+    gaussians_per_pixel: 1
+    gaussian_adapter:
+      gaussian_scale_min: 0.5
+      gaussian_scale_max: 15.0
+      sh_degree: 4
+    d_feature: 128
+    visualizer:
+      num_samples: 8
+      min_resolution: 256
+      export_ply: false
+    apply_bounds_shim: true
+    gs_params_head_type: dpt_gs
+    pose_free: true
+    pretrained_weights: ''
+  decoder:
+    name: splatting_cuda
+    background_color:
+    - 0.0
+    - 0.0
+    - 0.0
+    make_scale_invariant: false
+    use_gsplat: true
+  vggt:
+    input_image_shape:
+    - 518
+    - 518
+    head_mode: pcd
+    same_head: false
+    camera_prompt: false
+    num_visual_prompt: 0
+    use_extrinsic_token: false
+    use_intrinsic_token: false
+    detr:
+      name: deformable_detr
+      features_num_channel:
+      - 128
+      - 32
+      hidden_dim: 32
+      dim_down_factor: 1
+      dim_feedforward_factor: 4
+      enc_layers: 6
+      dec_layers: 1
+      dropout: 0.1
+      num_feature_levels: 2
+      dec_n_points: 4
+      enc_n_points: 4
+      split_threshold: 1.0
+      num_queries: 256
+      base_HW: 64
+      position_embedding: sine
+      pre_norm: false
+      self_layer_num: 1
+      num_level: 3
+      debug_mode: 0
+      decoder_type: conv
+      split_threshold_range:
+      - 0.0
+      - 2.0
+      only_last_level: false
+      grad_mode: absgrad
+      same_split_head: false
+      use_mean_features: true
+loss:
+  mse:
+    weight: 1.0
+  lpips:
+    weight: 0.05
+    apply_after_step: 0
+wandb:
+  project: noposplat
+  entity: scene-representation-group
+  name: re10k_hpc
+  mode: online
+  tags:
+  - re10k
+  - 256x256
+mode: train
+data_loader:
+  train:
+    num_workers: 16
+    persistent_workers: true
+    batch_size: 16
+    seed: 1234
+  test:
+    num_workers: 4
+    persistent_workers: false
+    batch_size: 1
+    seed: 2345
+  val:
+    num_workers: 1
+    persistent_workers: true
+    batch_size: 1
+    seed: 3456
+optimizer:
+  lr: 0.0002
+  warm_up_steps: 125
+  backbone_lr_multiplier: 0.1
+  backbone_trainable: GB+H
+  accumulate: 1
+checkpointing:
+  load: null
+  every_n_train_steps: 9375
+  save_top_k: 1
+  save_weights_only: true
+train:
+  depth_mode: null
+  extended_visualization: false
+  print_log_every_n_steps: 10
+  distiller: ''
+  distill_max_steps: 1000000
+  debug_mode: 0
+  camera_consis_loss: 0.0
+  camera_loss: 1.0
+  aux_loss: 0.0
+  one_sample_train: false
+  target_scale_mode: 2
+  extrinsic_drop: 0.1
+  intrinsic_drop: 0.1
+  align_corners: false
+  backbone_checkpoint: none
+  intrinsic_scaling: true
+  return_level_gs: true
+  train_context: false
+  use_pred_point_focal: false
+  scaling_detach: true
+  split_score_log_scale: true
+  split_score_grad_scale: 10000.0
+  split_loss: 0.01
+  random_split_strategy: score
+test:
+  output_path: test
+  align_pose: false
+  pose_align_steps: 100
+  rot_opt_lr: 0.005
+  trans_opt_lr: 0.005
+  compute_scores: true
+  save_image: false
+  save_video: false
+  save_compare: false
+  split_threshold_list:
+  - 0.0
+  - 0.1
+  - 0.2
+  - 0.3
+  - 0.4
+  pred_intrinsic: false
+seed: 111123
+trainer:
+  max_steps: 18751
+  val_check_interval: 500
+  gradient_clip_val: 0.5
+  num_nodes: 1
+dataset:
+  re10k:
+    make_baseline_1: true
+    relative_pose: true
+    augment: true
+    background_color:
+    - 0.0
+    - 0.0
+    - 0.0
+    overfit_to_scene: null
+    skip_bad_shape: true
+    view_sampler:
+      name: bounded
+      num_target_views: 4
+      num_context_views: 2
+      min_distance_between_context_views: 45
+      max_distance_between_context_views: 90
+      min_distance_to_context_views: 0
+      warm_up_steps: 9375
+      initial_min_distance_between_context_views: 25
+      initial_max_distance_between_context_views: 25
+      distill_steps: 0
+    name: re10k
+    roots:
+    - datasets/re10k
+    input_image_shape:
+    - 256
+    - 256
+    original_image_shape:
+    - 360
+    - 640
+    cameras_are_circular: false
+    baseline_min: 0.001
+    baseline_max: 10000000000.0
+    max_fov: 100.0

.hydra/hydra.yaml

@@ -0,0 +1,164 @@
+hydra:
+  run:
+    dir: outputs/${wandb.name}
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.mode=RUN
+    task:
+    - +experiment=re10k
+    - wandb.mode=online
+    - wandb.name=re10k_hpc
+  job:
+    name: main
+    chdir: null
+    override_dirname: +experiment=re10k,wandb.mode=online,wandb.name=re10k_hpc
+    id: ???
+    num: ???
+    config_name: main
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /root/highspeedstorage/injae/code/ICLR2026_HCP
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /root/highspeedstorage/injae/code/ICLR2026_HCP/config
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc
+    choices:
+      experiment: re10k
+      dataset@dataset.re10k: re10k
+      dataset/view_sampler_dataset_specific_config@dataset.re10k.view_sampler: bounded_re10k
+      dataset/view_sampler@dataset.re10k.view_sampler: bounded
+      model/decoder: splatting_cuda
+      model/encoder: noposplat
+      model/encoder/backbone: croco
+      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 +1,3 @@
+- +experiment=re10k
+- wandb.mode=online
+- wandb.name=re10k_hpc

main.log

@@ -0,0 +1,262 @@
+[2025-09-16 17:58:02,195][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 17:58:08,747][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 17:58:08,748][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 17:58:50,488][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 17:58:56,811][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 17:58:56,812][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:00:28,507][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:00:34,706][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:00:34,708][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:00:39,175][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=279` in the `DataLoader` to improve performance.
+
+[2025-09-16 18:00:41,384][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:01:31,562][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/utilities/data.py:79: Trying to infer the `batch_size` from an ambiguous collection. The batch size we found is 1. To avoid any miscalculations, use `self.log(..., batch_size=batch_size)`.
+
+[2025-09-16 18:01:31,563][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:01:31,564][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:01:33,246][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/functional.py:554: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /pytorch/aten/src/ATen/native/TensorShape.cpp:4314.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+
+[2025-09-16 18:02:35,041][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:02:41,051][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:02:41,052][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:04:06,193][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:04:11,843][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:04:11,844][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:04:15,683][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=279` in the `DataLoader` to improve performance.
+
+[2025-09-16 18:04:17,971][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:04:18,161][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/utilities/data.py:79: Trying to infer the `batch_size` from an ambiguous collection. The batch size we found is 1. To avoid any miscalculations, use `self.log(..., batch_size=batch_size)`.
+
+[2025-09-16 18:04:18,163][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:04:18,164][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:04:19,828][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:04:19,836][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/functional.py:554: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /pytorch/aten/src/ATen/native/TensorShape.cpp:4314.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+
+[2025-09-16 18:04:25,651][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:04:25,651][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:04:29,494][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=279` in the `DataLoader` to improve performance.
+
+[2025-09-16 18:04:42,154][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:04:48,178][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:04:48,179][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:04:52,770][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:04:52,934][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=279` in the `DataLoader` to improve performance.
+
+[2025-09-16 18:04:55,463][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:04:55,703][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/utilities/data.py:79: Trying to infer the `batch_size` from an ambiguous collection. The batch size we found is 1. To avoid any miscalculations, use `self.log(..., batch_size=batch_size)`.
+
+[2025-09-16 18:04:55,705][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:04:55,706][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:04:57,296][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/functional.py:554: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /pytorch/aten/src/ATen/native/TensorShape.cpp:4314.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+
+[2025-09-16 18:04:58,729][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:04:58,730][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:05:03,198][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=279` in the `DataLoader` to improve performance.
+
+[2025-09-16 18:05:05,392][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:05:05,581][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/utilities/data.py:79: Trying to infer the `batch_size` from an ambiguous collection. The batch size we found is 1. To avoid any miscalculations, use `self.log(..., batch_size=batch_size)`.
+
+[2025-09-16 18:05:05,583][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:05:05,583][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:05:07,261][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/functional.py:554: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /pytorch/aten/src/ATen/native/TensorShape.cpp:4314.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+
+[2025-09-16 18:08:24,727][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:08:30,965][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:08:30,966][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:14:16,577][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:14:22,737][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:14:22,738][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:16:48,200][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=34` in the `DataLoader` to improve performance.
+
+[2025-09-16 18:16:48,201][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 18:16:51,169][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:16:51,447][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/utilities/data.py:79: Trying to infer the `batch_size` from an ambiguous collection. The batch size we found is 1. To avoid any miscalculations, use `self.log(..., batch_size=batch_size)`.
+
+[2025-09-16 18:16:51,448][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:16:51,449][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:16:53,384][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/functional.py:554: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /pytorch/aten/src/ATen/native/TensorShape.cpp:4314.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+
+[2025-09-16 18:16:53,735][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/psnr', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-16 18:16:53,738][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/lpips', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-16 18:16:53,738][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/ssim', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-16 18:16:53,739][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/gaussian_num_ratio', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-16 18:17:04,137][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 18:23:12,050][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-16 19:44:06,679][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 19:44:14,557][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 19:44:14,557][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 19:45:55,937][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=34` in the `DataLoader` to improve performance.
+
+[2025-09-16 19:45:55,947][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 19:46:04,605][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 19:46:05,884][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/utilities/data.py:79: Trying to infer the `batch_size` from an ambiguous collection. The batch size we found is 1. To avoid any miscalculations, use `self.log(..., batch_size=batch_size)`.
+
+[2025-09-16 19:46:05,885][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 19:46:05,886][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 19:46:07,681][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/functional.py:554: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /pytorch/aten/src/ATen/native/TensorShape.cpp:4314.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+
+[2025-09-16 19:46:08,031][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/psnr', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-16 19:46:08,033][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/lpips', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-16 19:46:08,033][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/ssim', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-16 19:46:08,034][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/gaussian_num_ratio', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-16 19:46:18,197][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 19:52:10,048][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-17 08:15:57,790][dinov2][INFO] - using MLP layer as FFN
+[2025-09-17 08:16:05,563][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-17 08:16:05,564][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-17 08:17:32,793][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=34` in the `DataLoader` to improve performance.
+
+[2025-09-17 08:17:32,795][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-17 08:17:35,728][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-17 08:17:35,970][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/utilities/data.py:79: Trying to infer the `batch_size` from an ambiguous collection. The batch size we found is 1. To avoid any miscalculations, use `self.log(..., batch_size=batch_size)`.
+
+[2025-09-17 08:17:35,971][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-17 08:17:35,972][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-17 08:17:37,549][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/functional.py:554: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /pytorch/aten/src/ATen/native/TensorShape.cpp:4314.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+
+[2025-09-17 08:17:37,877][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/psnr', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-17 08:17:37,879][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/lpips', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-17 08:17:37,879][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/ssim', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-17 08:17:37,879][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('val/gaussian_num_ratio', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+
+[2025-09-17 08:17:47,740][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-17 08:23:38,721][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+

train_ddp_process_1.log

@@ -0,0 +1,81 @@
+[2025-09-16 17:58:19,330][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:02:49,892][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:03:08,637][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:03:08,638][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:08:41,358][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:09:17,206][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:09:17,206][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:14:31,602][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:14:50,266][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:14:50,267][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:16:48,201][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 18:17:00,737][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:17:04,136][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 18:23:12,050][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-16 19:44:23,040][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 19:44:42,938][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 19:44:42,939][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 19:45:55,947][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 19:46:15,532][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 19:46:18,197][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 19:52:10,048][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-17 08:16:14,871][dinov2][INFO] - using MLP layer as FFN
+[2025-09-17 08:16:47,663][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-17 08:16:47,664][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-17 08:17:32,794][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-17 08:17:44,025][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-17 08:17:47,739][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-17 08:23:38,705][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+

train_ddp_process_2.log

@@ -0,0 +1,81 @@
+[2025-09-16 17:58:18,234][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:02:50,404][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:03:28,788][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:03:28,793][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:08:40,323][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:08:59,994][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:08:59,995][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:14:32,040][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:14:51,075][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:14:51,077][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:16:48,201][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 18:17:00,237][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:17:04,136][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 18:23:12,054][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-16 19:44:22,989][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 19:44:41,440][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 19:44:41,441][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 19:45:55,947][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 19:46:14,712][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 19:46:18,197][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 19:52:10,049][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-17 08:16:15,147][dinov2][INFO] - using MLP layer as FFN
+[2025-09-17 08:16:47,824][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-17 08:16:47,825][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-17 08:17:32,795][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-17 08:17:44,084][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-17 08:17:47,739][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-17 08:23:38,704][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+

train_ddp_process_3.log

@@ -0,0 +1,81 @@
+[2025-09-16 17:58:19,053][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:02:50,349][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:03:29,558][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:03:29,561][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:08:41,428][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:09:17,198][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:09:17,199][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:14:32,923][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:15:09,086][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:15:09,087][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:16:48,201][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 18:17:00,926][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:17:04,134][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 18:23:12,047][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-16 19:44:23,779][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 19:44:55,897][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 19:44:55,899][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 19:45:55,947][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 19:46:15,252][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 19:46:18,197][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 19:52:10,048][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-17 08:16:14,263][dinov2][INFO] - using MLP layer as FFN
+[2025-09-17 08:16:34,545][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-17 08:16:34,546][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-17 08:17:32,795][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-17 08:17:43,969][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-17 08:17:47,738][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-17 08:23:38,705][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+

train_ddp_process_4.log

@@ -0,0 +1,81 @@
+[2025-09-16 17:58:18,473][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:02:50,329][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:03:27,530][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:03:27,531][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:08:41,389][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:09:17,146][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:09:17,147][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:14:32,867][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:15:09,006][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:15:09,007][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:16:48,201][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 18:17:01,226][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:17:04,135][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 18:23:12,055][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-16 19:44:23,799][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 19:44:56,745][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 19:44:56,746][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 19:45:55,947][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 19:46:14,882][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 19:46:18,197][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 19:52:10,048][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-17 08:16:14,950][dinov2][INFO] - using MLP layer as FFN
+[2025-09-17 08:16:47,187][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-17 08:16:47,188][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-17 08:17:32,794][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-17 08:17:44,542][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-17 08:17:47,739][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-17 08:23:38,710][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+

train_ddp_process_5.log

@@ -0,0 +1,81 @@
+[2025-09-16 17:58:18,567][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:02:51,037][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:03:30,313][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:03:30,315][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:08:39,963][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:08:59,630][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:08:59,630][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:14:32,775][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:15:08,948][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:15:08,957][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:16:48,200][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 18:17:00,467][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:17:04,134][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 18:23:12,042][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-16 19:44:23,247][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 19:44:43,991][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 19:44:43,991][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 19:45:55,947][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 19:46:15,452][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 19:46:18,197][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 19:52:10,048][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-17 08:16:13,809][dinov2][INFO] - using MLP layer as FFN
+[2025-09-17 08:16:33,675][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-17 08:16:33,676][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-17 08:17:32,794][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-17 08:17:44,220][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-17 08:17:47,736][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-17 08:23:38,709][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+

train_ddp_process_6.log

@@ -0,0 +1,81 @@
+[2025-09-16 17:58:19,111][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:02:51,004][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:03:30,361][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:03:30,363][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:08:40,776][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:09:13,164][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:09:13,165][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:14:32,593][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:15:06,668][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:15:06,673][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:16:48,201][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 18:17:01,180][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:17:04,136][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 18:23:12,046][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-16 19:44:23,838][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 19:44:56,721][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 19:44:56,722][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 19:45:55,947][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 19:46:15,168][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 19:46:18,197][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 19:52:10,048][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-17 08:16:14,202][dinov2][INFO] - using MLP layer as FFN
+[2025-09-17 08:16:34,429][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-17 08:16:34,429][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-17 08:17:32,794][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-17 08:17:44,160][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-17 08:17:47,738][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-17 08:23:38,716][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+

train_ddp_process_7.log

@@ -0,0 +1,81 @@
+[2025-09-16 17:58:18,319][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:02:51,163][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:03:30,264][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:03:30,265][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:08:40,815][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:09:16,606][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:09:16,607][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:14:32,775][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 18:15:08,930][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:15:08,931][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 18:16:48,201][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 18:17:00,823][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:17:04,135][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 18:23:12,046][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-16 19:44:23,886][dinov2][INFO] - using MLP layer as FFN
+[2025-09-16 19:44:56,838][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 19:44:56,839][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-16 19:45:55,947][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-16 19:46:15,492][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 19:46:18,197][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-16 19:52:10,048][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+
+[2025-09-17 08:16:14,915][dinov2][INFO] - using MLP layer as FFN
+[2025-09-17 08:16:47,804][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-17 08:16:47,805][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+[2025-09-17 08:17:32,794][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. 
+  warnings.warn(  # warn only once
+
+[2025-09-17 08:17:45,024][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation. 
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-17 08:17:47,738][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/autograd/graph.py:824: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
+grad.sizes() = [56, 128, 1, 1], strides() = [128, 1, 128, 128]
+bucket_view.sizes() = [56, 128, 1, 1], strides() = [128, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:328.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+
+[2025-09-17 08:23:38,704][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:198: UserWarning: The epoch parameter in `scheduler.step()` was not necessary and is being deprecated where possible. Please use `scheduler.step()` to step the scheduler. During the deprecation, if epoch is different from None, the closed form is used instead of the new chainable form, where available. Please open an issue if you are unable to replicate your use case: https://github.com/pytorch/pytorch/issues/new/choose.
+  warnings.warn(EPOCH_DEPRECATION_WARNING, UserWarning)
+

wandb/debug-internal.log

@@ -0,0 +1,7 @@
+{"time":"2025-09-17T08:17:30.716918246+08:00","level":"INFO","msg":"stream: starting","core version":"0.19.9","symlink path":"/root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc/wandb/run-20250917_081730-bcwwe2gj/logs/debug-core.log"}
+{"time":"2025-09-17T08:17:31.176856401+08:00","level":"INFO","msg":"created new stream","id":"bcwwe2gj"}
+{"time":"2025-09-17T08:17:31.177025464+08:00","level":"INFO","msg":"stream: started","id":"bcwwe2gj"}
+{"time":"2025-09-17T08:17:31.177197947+08:00","level":"INFO","msg":"writer: Do: started","stream_id":"bcwwe2gj"}
+{"time":"2025-09-17T08:17:31.177369747+08:00","level":"INFO","msg":"handler: started","stream_id":"bcwwe2gj"}
+{"time":"2025-09-17T08:17:31.177482294+08:00","level":"INFO","msg":"sender: started","stream_id":"bcwwe2gj"}
+{"time":"2025-09-17T08:17:31.468144818+08:00","level":"INFO","msg":"Starting system monitor"}

wandb/debug.log

@@ -0,0 +1,22 @@
+2025-09-17 08:17:30,694 INFO    MainThread:31741 [wandb_setup.py:_flush():67] Current SDK version is 0.19.9
+2025-09-17 08:17:30,694 INFO    MainThread:31741 [wandb_setup.py:_flush():67] Configure stats pid to 31741
+2025-09-17 08:17:30,694 INFO    MainThread:31741 [wandb_setup.py:_flush():67] Loading settings from /root/.config/wandb/settings
+2025-09-17 08:17:30,694 INFO    MainThread:31741 [wandb_setup.py:_flush():67] Loading settings from /root/highspeedstorage/injae/code/ICLR2026_HCP/wandb/settings
+2025-09-17 08:17:30,694 INFO    MainThread:31741 [wandb_setup.py:_flush():67] Loading settings from environment variables
+2025-09-17 08:17:30,694 INFO    MainThread:31741 [wandb_init.py:setup_run_log_directory():662] Logging user logs to /root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc/wandb/run-20250917_081730-bcwwe2gj/logs/debug.log
+2025-09-17 08:17:30,695 INFO    MainThread:31741 [wandb_init.py:setup_run_log_directory():663] Logging internal logs to /root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc/wandb/run-20250917_081730-bcwwe2gj/logs/debug-internal.log
+2025-09-17 08:17:30,695 INFO    MainThread:31741 [wandb_init.py:init():781] calling init triggers
+2025-09-17 08:17:30,695 INFO    MainThread:31741 [wandb_init.py:init():786] wandb.init called with sweep_config: {}
+config: {'model': {'encoder': {'backbone': {'name': 'croco', 'model': 'ViTLarge_BaseDecoder', 'patch_embed_cls': 'PatchEmbedDust3R', 'asymmetry_decoder': True, 'intrinsics_embed_loc': 'encoder', 'intrinsics_embed_degree': 4, 'intrinsics_embed_type': 'token'}, 'name': 'noposplat', 'opacity_mapping': {'initial': 0.0, 'final': 0.0, 'warm_up': 1}, 'num_monocular_samples': 32, 'num_surfaces': 1, 'predict_opacity': False, 'gaussians_per_pixel': 1, 'gaussian_adapter': {'gaussian_scale_min': 0.5, 'gaussian_scale_max': 15.0, 'sh_degree': 4}, 'd_feature': 128, 'visualizer': {'num_samples': 8, 'min_resolution': 256, 'export_ply': False}, 'apply_bounds_shim': True, 'gs_params_head_type': 'dpt_gs', 'pose_free': True, 'pretrained_weights': ''}, 'decoder': {'name': 'splatting_cuda', 'background_color': [0.0, 0.0, 0.0], 'make_scale_invariant': False, 'use_gsplat': True}, 'vggt': {'input_image_shape': [518, 518], 'head_mode': 'pcd', 'same_head': False, 'camera_prompt': False, 'num_visual_prompt': 0, 'use_extrinsic_token': False, 'use_intrinsic_token': False, 'detr': {'name': 'deformable_detr', 'features_num_channel': [128, 32], 'hidden_dim': 32, 'dim_down_factor': 1, 'dim_feedforward_factor': 4, 'enc_layers': 6, 'dec_layers': 1, 'dropout': 0.1, 'num_feature_levels': 2, 'dec_n_points': 4, 'enc_n_points': 4, 'split_threshold': 1.0, 'num_queries': 256, 'base_HW': 64, 'position_embedding': 'sine', 'pre_norm': False, 'self_layer_num': 1, 'num_level': 3, 'debug_mode': 0, 'decoder_type': 'conv', 'split_threshold_range': [0.0, 2.0], 'only_last_level': False, 'grad_mode': 'absgrad', 'same_split_head': False, 'use_mean_features': True}}}, 'loss': {'mse': {'weight': 1.0}, 'lpips': {'weight': 0.05, 'apply_after_step': 0}}, 'wandb': {'project': 'noposplat', 'entity': 'scene-representation-group', 'name': 're10k_hpc', 'mode': 'online', 'tags': ['re10k', '256x256']}, 'mode': 'train', 'data_loader': {'train': {'num_workers': 16, 'persistent_workers': True, 'batch_size': 16, 'seed': 1234}, 'test': {'num_workers': 4, 'persistent_workers': False, 'batch_size': 1, 'seed': 2345}, 'val': {'num_workers': 1, 'persistent_workers': True, 'batch_size': 1, 'seed': 3456}}, 'optimizer': {'lr': 0.0002, 'warm_up_steps': 125, 'backbone_lr_multiplier': 0.1, 'backbone_trainable': 'GB+H', 'accumulate': 1}, 'checkpointing': {'load': None, 'every_n_train_steps': 9375, 'save_top_k': 1, 'save_weights_only': True}, 'train': {'depth_mode': None, 'extended_visualization': False, 'print_log_every_n_steps': 10, 'distiller': '', 'distill_max_steps': 1000000, 'debug_mode': 0, 'camera_consis_loss': 0.0, 'camera_loss': 1.0, 'aux_loss': 0.0, 'one_sample_train': False, 'target_scale_mode': 2, 'extrinsic_drop': 0.1, 'intrinsic_drop': 0.1, 'align_corners': False, 'backbone_checkpoint': 'none', 'intrinsic_scaling': True, 'return_level_gs': True, 'train_context': False, 'use_pred_point_focal': False, 'scaling_detach': True, 'split_score_log_scale': True, 'split_score_grad_scale': 10000.0, 'split_loss': 0.01, 'random_split_strategy': 'score'}, 'test': {'output_path': 'test', 'align_pose': False, 'pose_align_steps': 100, 'rot_opt_lr': 0.005, 'trans_opt_lr': 0.005, 'compute_scores': True, 'save_image': False, 'save_video': False, 'save_compare': False, 'split_threshold_list': [0.0, 0.1, 0.2, 0.3, 0.4], 'pred_intrinsic': False}, 'seed': 111123, 'trainer': {'max_steps': 18751, 'val_check_interval': 500, 'gradient_clip_val': 0.5, 'num_nodes': 1}, 'dataset': {'re10k': {'make_baseline_1': True, 'relative_pose': True, 'augment': True, 'background_color': [0.0, 0.0, 0.0], 'overfit_to_scene': None, 'skip_bad_shape': True, 'view_sampler': {'name': 'bounded', 'num_target_views': 4, 'num_context_views': 2, 'min_distance_between_context_views': 45, 'max_distance_between_context_views': 90, 'min_distance_to_context_views': 0, 'warm_up_steps': 9375, 'initial_min_distance_between_context_views': 25, 'initial_max_distance_between_context_views': 25, 'distill_steps': 0}, 'name': 're10k', 'roots': ['datasets/re10k'], 'input_image_shape': [256, 256], 'original_image_shape': [360, 640], 'cameras_are_circular': False, 'baseline_min': 0.001, 'baseline_max': 10000000000.0, 'max_fov': 100.0}}, '_wandb': {}}
+2025-09-17 08:17:30,695 INFO    MainThread:31741 [wandb_init.py:init():809] starting backend
+2025-09-17 08:17:30,695 INFO    MainThread:31741 [wandb_init.py:init():813] sending inform_init request
+2025-09-17 08:17:30,711 INFO    MainThread:31741 [backend.py:_multiprocessing_setup():101] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2025-09-17 08:17:30,711 INFO    MainThread:31741 [wandb_init.py:init():823] backend started and connected
+2025-09-17 08:17:30,725 INFO    MainThread:31741 [wandb_init.py:init():915] updated telemetry
+2025-09-17 08:17:30,760 INFO    MainThread:31741 [wandb_init.py:init():939] communicating run to backend with 90.0 second timeout
+2025-09-17 08:17:31,459 INFO    MainThread:31741 [wandb_init.py:init():1014] starting run threads in backend
+2025-09-17 08:17:31,629 INFO    MainThread:31741 [wandb_run.py:_console_start():2454] atexit reg
+2025-09-17 08:17:31,630 INFO    MainThread:31741 [wandb_run.py:_redirect():2306] redirect: wrap_raw
+2025-09-17 08:17:31,632 INFO    MainThread:31741 [wandb_run.py:_redirect():2371] Wrapping output streams.
+2025-09-17 08:17:31,632 INFO    MainThread:31741 [wandb_run.py:_redirect():2394] Redirects installed.
+2025-09-17 08:17:31,635 INFO    MainThread:31741 [wandb_init.py:init():1056] run started, returning control to user process

wandb/run-20250916_180036-llem5odx/files/config.yaml

@@ -0,0 +1,423 @@
+_wandb:
+    value:
+        cli_version: 0.19.9
+        m:
+            - "1": points_context_imgs.height
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": trainer/global_step
+              "6":
+                - 3
+              "7": []
+            - "1": comparison.count
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.format
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs._type
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.count
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.height
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.captions
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison.width
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.count
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.filenames
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison.filenames
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison.format
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.filenames
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison._type
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison.height
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.width
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.captions
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.width
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.format
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.width
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison.captions
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.count
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.filenames
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.captions
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.height
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.format
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization._type
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map._type
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+        python_version: 3.12.11
+        t:
+            "1":
+                - 1
+                - 41
+                - 49
+                - 50
+                - 55
+                - 106
+            "2":
+                - 1
+                - 41
+                - 49
+                - 50
+                - 55
+                - 106
+            "3":
+                - 7
+                - 13
+                - 15
+                - 16
+                - 23
+                - 55
+                - 66
+            "4": 3.12.11
+            "5": 0.19.9
+            "8":
+                - 5
+            "12": 0.19.9
+            "13": linux-x86_64
+checkpointing:
+    value:
+        every_n_train_steps: 9375
+        load: null
+        save_top_k: 1
+        save_weights_only: true
+data_loader:
+    value:
+        test:
+            batch_size: 1
+            num_workers: 4
+            persistent_workers: false
+            seed: 2345
+        train:
+            batch_size: 16
+            num_workers: 16
+            persistent_workers: true
+            seed: 1234
+        val:
+            batch_size: 1
+            num_workers: 1
+            persistent_workers: true
+            seed: 3456
+dataset:
+    value:
+        re10k:
+            augment: true
+            background_color:
+                - 0
+                - 0
+                - 0
+            baseline_max: 1e+10
+            baseline_min: 0.001
+            cameras_are_circular: false
+            input_image_shape:
+                - 256
+                - 256
+            make_baseline_1: true
+            max_fov: 100
+            name: re10k
+            original_image_shape:
+                - 360
+                - 640
+            overfit_to_scene: null
+            relative_pose: true
+            roots:
+                - datasets/re10k
+            skip_bad_shape: true
+            view_sampler:
+                distill_steps: 0
+                initial_max_distance_between_context_views: 25
+                initial_min_distance_between_context_views: 25
+                max_distance_between_context_views: 90
+                min_distance_between_context_views: 45
+                min_distance_to_context_views: 0
+                name: bounded
+                num_context_views: 2
+                num_target_views: 4
+                warm_up_steps: 9375
+loss:
+    value:
+        lpips:
+            apply_after_step: 0
+            weight: 0.05
+        mse:
+            weight: 1
+mode:
+    value: train
+model:
+    value:
+        decoder:
+            background_color:
+                - 0
+                - 0
+                - 0
+            make_scale_invariant: false
+            name: splatting_cuda
+            use_gsplat: true
+        encoder:
+            apply_bounds_shim: true
+            backbone:
+                asymmetry_decoder: true
+                intrinsics_embed_degree: 4
+                intrinsics_embed_loc: encoder
+                intrinsics_embed_type: token
+                model: ViTLarge_BaseDecoder
+                name: croco
+                patch_embed_cls: PatchEmbedDust3R
+            d_feature: 128
+            gaussian_adapter:
+                gaussian_scale_max: 15
+                gaussian_scale_min: 0.5
+                sh_degree: 4
+            gaussians_per_pixel: 1
+            gs_params_head_type: dpt_gs
+            name: noposplat
+            num_monocular_samples: 32
+            num_surfaces: 1
+            opacity_mapping:
+                final: 0
+                initial: 0
+                warm_up: 1
+            pose_free: true
+            predict_opacity: false
+            pretrained_weights: ""
+            visualizer:
+                export_ply: false
+                min_resolution: 256
+                num_samples: 8
+        vggt:
+            camera_prompt: false
+            detr:
+                base_HW: 64
+                debug_mode: 0
+                dec_layers: 1
+                dec_n_points: 4
+                decoder_type: conv
+                dim_down_factor: 1
+                dim_feedforward_factor: 4
+                dropout: 0.1
+                enc_layers: 6
+                enc_n_points: 4
+                features_num_channel:
+                    - 128
+                    - 32
+                grad_mode: absgrad
+                hidden_dim: 32
+                name: deformable_detr
+                num_feature_levels: 2
+                num_level: 3
+                num_queries: 256
+                only_last_level: false
+                position_embedding: sine
+                pre_norm: false
+                same_split_head: false
+                self_layer_num: 1
+                split_threshold: 1
+                split_threshold_range:
+                    - 0
+                    - 2
+                use_mean_features: true
+            head_mode: pcd
+            input_image_shape:
+                - 518
+                - 518
+            num_visual_prompt: 0
+            same_head: false
+            use_extrinsic_token: false
+            use_intrinsic_token: false
+optimizer:
+    value:
+        accumulate: 1
+        backbone_lr_multiplier: 0.1
+        backbone_trainable: GB+H
+        lr: 0.0002
+        warm_up_steps: 125
+seed:
+    value: 111123
+test:
+    value:
+        align_pose: false
+        compute_scores: true
+        output_path: test
+        pose_align_steps: 100
+        pred_intrinsic: false
+        rot_opt_lr: 0.005
+        save_compare: false
+        save_image: false
+        save_video: false
+        split_threshold_list:
+            - 0
+            - 0.1
+            - 0.2
+            - 0.3
+            - 0.4
+        trans_opt_lr: 0.005
+train:
+    value:
+        align_corners: false
+        aux_loss: 0
+        backbone_checkpoint: none
+        camera_consis_loss: 0
+        camera_loss: 1
+        debug_mode: 0
+        depth_mode: null
+        distill_max_steps: 1000000
+        distiller: ""
+        extended_visualization: false
+        extrinsic_drop: 0.1
+        intrinsic_drop: 0.1
+        intrinsic_scaling: true
+        one_sample_train: false
+        print_log_every_n_steps: 10
+        random_split_strategy: score
+        return_level_gs: true
+        scaling_detach: true
+        split_loss: 0.01
+        split_score_grad_scale: 10000
+        split_score_log_scale: true
+        target_scale_mode: 2
+        train_context: false
+        use_pred_point_focal: false
+trainer:
+    value:
+        gradient_clip_val: 0.5
+        max_steps: 18751
+        num_nodes: 1
+        val_check_interval: 500
+wandb:
+    value:
+        entity: scene-representation-group
+        mode: online
+        name: re10k_hpc
+        project: noposplat
+        tags:
+            - re10k
+            - 256x256

wandb/run-20250916_180036-llem5odx/files/output.log

@@ -0,0 +1,40 @@
+LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0]
+
+  | Name    | Type                 | Params | Mode
+---------------------------------------------------------
+0 | vggt    | VGGT                 | 585 M  | train
+1 | decoder | DecoderSplattingCUDA | 0      | train
+2 | losses  | ModuleList           | 0      | train
+---------------------------------------------------------
+585 M     Trainable params
+0         Non-trainable params
+585 M     Total params
+2,340.472 Total estimated model params size (MB)
+773       Modules in train mode
+980       Modules in eval mode
+Sanity Checking: |                                                                                                                                                                                                    | 0/? [00:00<?, ?it/s][2025-09-16 18:00:39,175][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=279` in the `DataLoader` to improve performance.
+
+Validation epoch start on rank 0
+Sanity Checking DataLoader 0:   0%|                                                                                                                                                                                   | 0/1 [00:00<?, ?it/s]validation step 0; scene = ['306e2b7785657539']; context = [[48, 73]]
+target intrinsic:  tensor(0.8595, device='cuda:0') tensor(0.8597, device='cuda:0')
+pred intrinsic:  tensor(0.8780, device='cuda:0') tensor(0.8776, device='cuda:0')
+[2025-09-16 18:00:41,384][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation.
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:01:31,562][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/utilities/data.py:79: Trying to infer the `batch_size` from an ambiguous collection. The batch size we found is 1. To avoid any miscalculations, use `self.log(..., batch_size=batch_size)`.
+
+Setting up [LPIPS] perceptual loss: trunk [vgg], v[0.1], spatial [off]
+[2025-09-16 18:01:31,563][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:01:31,564][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+Loading model from: /opt/conda/envs/ours/lib/python3.12/site-packages/lpips/weights/v0.1/vgg.pth
+[2025-09-16 18:01:33,246][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/functional.py:554: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /pytorch/aten/src/ATen/native/TensorShape.cpp:4314.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+
+Epoch 0: |                                                                                                                                                                                        | 9/? [00:26<00:00,  0.34it/s, v_num=5odx]
+
+Detected KeyboardInterrupt, attempting graceful shutdown ...

wandb/run-20250916_180036-llem5odx/files/requirements.txt

@@ -0,0 +1,129 @@
+ruff==0.11.4
+PyYAML==6.0.2
+sympy==1.13.3
+fsspec==2024.6.1
+imageio-ffmpeg==0.6.0
+setuptools==78.1.1
+timm==1.0.15
+fonttools==4.57.0
+nvidia-cublas-cu12==12.8.3.14
+nvidia-cuda-nvrtc-cu12==12.8.61
+dacite==1.9.2
+hydra-core==1.3.2
+certifi==2022.12.7
+rich==14.0.0
+filelock==3.13.1
+idna==3.4
+nvidia-cuda-cupti-cu12==12.8.57
+pip==25.2
+wandb==0.19.9
+ninja==1.11.1.4
+networkx==3.3
+GitPython==3.1.44
+moviepy==1.0.3
+markdown-it-py==3.0.0
+lightning==2.5.1
+Jinja2==3.1.4
+nvidia-cusparse-cu12==12.5.7.53
+tabulate==0.9.0
+tifffile==2025.3.30
+nvidia-nvtx-cu12==12.8.55
+jaxtyping==0.2.37
+aiohappyeyeballs==2.6.1
+MarkupSafe==2.1.5
+pydantic_core==2.33.1
+mypy-extensions==1.0.0
+lazy_loader==0.4
+pydantic==2.11.2
+typing-inspection==0.4.0
+pillow==10.4.0
+nvidia-nvjitlink-cu12==12.8.61
+omegaconf==2.3.0
+scikit-image==0.25.2
+jaraco.context==5.3.0
+typing_extensions==4.12.2
+mpmath==1.3.0
+opencv-python==4.11.0.86
+backports.tarfile==1.2.0
+async-timeout==5.0.1
+opt_einsum==3.4.0
+nvidia-cuda-runtime-cu12==12.8.57
+numpy==1.26.4
+torch==2.7.1+cu128
+psutil==7.0.0
+pytorch-lightning==2.5.1
+protobuf==5.29.4
+imageio==2.37.0
+colorama==0.4.6
+attrs==25.3.0
+importlib_metadata==8.0.0
+frozenlist==1.5.0
+beartype==0.19.0
+propcache==0.3.1
+typeguard==4.3.0
+torchaudio==2.7.1+cu128
+decorator==4.4.2
+nvidia-cudnn-cu12==9.7.1.26
+torchvision==0.22.1+cu128
+sentry-sdk==2.25.1
+mdurl==0.1.2
+more-itertools==10.3.0
+einops==0.8.1
+zipp==3.19.2
+antlr4-python3-runtime==4.9.3
+cycler==0.12.1
+nvidia-cufile-cu12==1.13.0.11
+jaraco.collections==5.1.0
+jaraco.text==3.12.1
+torchmetrics==1.8.2
+packaging==24.2
+black==25.1.0
+setproctitle==1.3.5
+Pygments==2.19.2
+tomli==2.2.1
+six==1.17.0
+docker-pycreds==0.4.0
+inflect==7.3.1
+requests==2.28.1
+svg.py==1.6.0
+python-dateutil==2.9.0.post0
+proglog==0.1.11
+gsplat==1.5.3
+e3nn==0.5.6
+pathspec==0.12.1
+aiosignal==1.3.2
+wheel==0.45.1
+multidict==6.3.2
+pyparsing==3.2.3
+plyfile==1.1
+jaraco.functools==4.0.1
+nvidia-nccl-cu12==2.26.2
+triton==3.3.1
+nvidia-cusparselt-cu12==0.6.3
+scikit-video==1.1.11
+huggingface-hub==0.30.1
+lightning-utilities==0.14.3
+wadler_lindig==0.1.4
+nvidia-curand-cu12==10.3.9.55
+matplotlib==3.10.1
+kiwisolver==1.4.8
+tqdm==4.67.1
+safetensors==0.5.3
+scipy==1.15.2
+lpips==0.1.4
+nvidia-cusolver-cu12==11.7.2.55
+click==8.1.8
+charset-normalizer==2.1.1
+platformdirs==4.3.7
+opt-einsum-fx==0.1.4
+annotated-types==0.7.0
+smmap==5.0.2
+gitdb==4.0.12
+nvidia-cufft-cu12==11.3.3.41
+contourpy==1.3.1
+urllib3==1.26.13
+autocommand==2.2.2
+yarl==1.19.0
+aiohttp==3.11.16
+colorspacious==1.1.2
+python-dotenv==1.1.0

wandb/run-20250916_180036-llem5odx/files/wandb-metadata.json

@@ -0,0 +1,79 @@
+{
+  "os": "Linux-5.15.0-153-generic-x86_64-with-glibc2.35",
+  "python": "CPython 3.12.11",
+  "startedAt": "2025-09-16T10:00:36.796366Z",
+  "args": [
+    "+experiment=re10k",
+    "wandb.mode=online",
+    "wandb.name=re10k_hpc"
+  ],
+  "program": "-m src.main",
+  "git": {
+    "remote": "git@github.com:K-nowing/ICLR2026_HCP.git",
+    "commit": "6c0080eae0295d43dd7f18fd71e742d924238f89"
+  },
+  "email": "dna9041@korea.ac.kr",
+  "root": "/root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc",
+  "host": "notebook-72e474dd-92e1-11f0-a6ea-7ab79f1c53cb-0",
+  "executable": "/opt/conda/envs/ours/bin/python",
+  "cpu_count": 144,
+  "cpu_count_logical": 288,
+  "gpu": "NVIDIA B200",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "1081101176832",
+      "used": "211821359104"
+    }
+  },
+  "memory": {
+    "total": "2434690400256"
+  },
+  "cpu": {
+    "count": 144,
+    "countLogical": 288
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    }
+  ],
+  "cudaVersion": "12.8"
+}

wandb/run-20250916_180036-llem5odx/files/wandb-summary.json

@@ -0,0 +1 @@
+{"activation_visualization":{"_type":"images/separated","width":256,"height":256,"format":"png","count":1,"filenames":["media/images/activation_visualization_2_e7083dae4b9eda33b264.png"],"captions":["306e2b7785657539"]},"_step":3,"comparison":{"_type":"images/separated","width":1064,"height":1098,"format":"png","count":1,"filenames":["media/images/comparison_0_621c28fb8c1f6ff0dbb4.png"],"captions":["306e2b7785657539"]},"_timestamp":1.7580168936113427e+09,"trainer/global_step":0,"points_context_imgs":{"height":800,"format":"png","count":1,"filenames":["media/images/points_context_imgs_1_cbb5017831dfea54f21d.png"],"captions":["306e2b7785657539"],"_type":"images/separated","width":536},"attention_map":{"width":800,"height":536,"format":"png","count":1,"filenames":["media/images/attention_map_3_4bf1d5f13f9f4f399591.png"],"captions":["306e2b7785657539"],"_type":"images/separated"},"_wandb":{"runtime":96},"_runtime":56.816239772}

wandb/run-20250916_180413-d4pobtwb/files/output.log

@@ -0,0 +1,83 @@
+LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [1]
+
+  | Name    | Type                 | Params | Mode
+---------------------------------------------------------
+0 | vggt    | VGGT                 | 585 M  | train
+1 | decoder | DecoderSplattingCUDA | 0      | train
+2 | losses  | ModuleList           | 0      | train
+---------------------------------------------------------
+585 M     Trainable params
+0         Non-trainable params
+585 M     Total params
+2,340.472 Total estimated model params size (MB)
+773       Modules in train mode
+980       Modules in eval mode
+Sanity Checking: |                                                                                                                                                                                                    | 0/? [00:00<?, ?it/s][2025-09-16 18:04:15,683][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=279` in the `DataLoader` to improve performance.
+
+Validation epoch start on rank 0
+Sanity Checking DataLoader 0:   0%|                                                                                                                                                                                   | 0/1 [00:00<?, ?it/s]validation step 0; scene = ['306e2b7785657539']; context = [[48, 73]]
+target intrinsic:  tensor(0.8595, device='cuda:0') tensor(0.8597, device='cuda:0')
+pred intrinsic:  tensor(0.8780, device='cuda:0') tensor(0.8776, device='cuda:0')
+[2025-09-16 18:04:17,971][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/utils/cpp_extension.py:2356: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation.
+If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
+  warnings.warn(
+
+[2025-09-16 18:04:18,161][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/lightning/pytorch/utilities/data.py:79: Trying to infer the `batch_size` from an ambiguous collection. The batch size we found is 1. To avoid any miscalculations, use `self.log(..., batch_size=batch_size)`.
+
+Setting up [LPIPS] perceptual loss: trunk [vgg], v[0.1], spatial [off]
+[2025-09-16 18:04:18,163][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
+  warnings.warn(
+
+[2025-09-16 18:04:18,164][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=VGG16_Weights.IMAGENET1K_V1`. You can also use `weights=VGG16_Weights.DEFAULT` to get the most up-to-date weights.
+  warnings.warn(msg)
+
+Loading model from: /opt/conda/envs/ours/lib/python3.12/site-packages/lpips/weights/v0.1/vgg.pth
+[2025-09-16 18:04:19,836][py.warnings][WARNING] - /opt/conda/envs/ours/lib/python3.12/site-packages/torch/functional.py:554: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /pytorch/aten/src/ATen/native/TensorShape.cpp:4314.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+
+Epoch 0: |                                                                                                                                                                                                            | 0/? [00:00<?, ?it/s]
+
+Detected KeyboardInterrupt, attempting graceful shutdown ...
+Exception ignored in: <module 'threading' from '/opt/conda/envs/ours/lib/python3.12/threading.py'>
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/threading.py", line 1575, in _shutdown
+    def _shutdown():
+
+KeyboardInterrupt:
+Exception ignored in atexit callback: <function _start_and_connect_service.<locals>.teardown_atexit at 0x7f90148e3060>
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/service_connection.py", line 94, in teardown_atexit
+    conn.teardown(hooks.exit_code)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/service_connection.py", line 226, in teardown
+    self._router.join()
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/router.py", line 75, in join
+    self._thread.join()
+  File "/opt/conda/envs/ours/lib/python3.12/threading.py", line 1149, in join
+    self._wait_for_tstate_lock()
+  File "/opt/conda/envs/ours/lib/python3.12/threading.py", line 1169, in _wait_for_tstate_lock
+    if lock.acquire(block, timeout):
+       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+KeyboardInterrupt:
+Exception ignored in atexit callback: <function _exit_function at 0x7f90d829e160>
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/multiprocessing/util.py", line 337, in _exit_function
+    _run_finalizers(0)
+  File "/opt/conda/envs/ours/lib/python3.12/multiprocessing/util.py", line 303, in _run_finalizers
+    finalizer()
+  File "/opt/conda/envs/ours/lib/python3.12/multiprocessing/util.py", line 227, in __call__
+    res = self._callback(*self._args, **self._kwargs)
+          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/multiprocessing/managers.py", line 675, in _finalize_manager
+    process.join(timeout=shutdown_timeout)
+  File "/opt/conda/envs/ours/lib/python3.12/multiprocessing/process.py", line 149, in join
+    res = self._popen.wait(timeout)
+          ^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/multiprocessing/popen_fork.py", line 40, in wait
+    if not wait([self.sentinel], timeout):
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/multiprocessing/connection.py", line 1136, in wait
+    ready = selector.select(timeout)
+            ^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/selectors.py", line 415, in select
+    fd_event_list = self._selector.poll(timeout)
+                    ^^^^^^^^^^^^^^KeyboardInterrupt:

wandb/run-20250916_180413-d4pobtwb/files/requirements.txt

@@ -0,0 +1,129 @@
+ruff==0.11.4
+PyYAML==6.0.2
+sympy==1.13.3
+fsspec==2024.6.1
+imageio-ffmpeg==0.6.0
+setuptools==78.1.1
+timm==1.0.15
+fonttools==4.57.0
+nvidia-cublas-cu12==12.8.3.14
+nvidia-cuda-nvrtc-cu12==12.8.61
+dacite==1.9.2
+hydra-core==1.3.2
+certifi==2022.12.7
+rich==14.0.0
+filelock==3.13.1
+idna==3.4
+nvidia-cuda-cupti-cu12==12.8.57
+pip==25.2
+wandb==0.19.9
+ninja==1.11.1.4
+networkx==3.3
+GitPython==3.1.44
+moviepy==1.0.3
+markdown-it-py==3.0.0
+lightning==2.5.1
+Jinja2==3.1.4
+nvidia-cusparse-cu12==12.5.7.53
+tabulate==0.9.0
+tifffile==2025.3.30
+nvidia-nvtx-cu12==12.8.55
+jaxtyping==0.2.37
+aiohappyeyeballs==2.6.1
+MarkupSafe==2.1.5
+pydantic_core==2.33.1
+mypy-extensions==1.0.0
+lazy_loader==0.4
+pydantic==2.11.2
+typing-inspection==0.4.0
+pillow==10.4.0
+nvidia-nvjitlink-cu12==12.8.61
+omegaconf==2.3.0
+scikit-image==0.25.2
+jaraco.context==5.3.0
+typing_extensions==4.12.2
+mpmath==1.3.0
+opencv-python==4.11.0.86
+backports.tarfile==1.2.0
+async-timeout==5.0.1
+opt_einsum==3.4.0
+nvidia-cuda-runtime-cu12==12.8.57
+numpy==1.26.4
+torch==2.7.1+cu128
+psutil==7.0.0
+pytorch-lightning==2.5.1
+protobuf==5.29.4
+imageio==2.37.0
+colorama==0.4.6
+attrs==25.3.0
+importlib_metadata==8.0.0
+frozenlist==1.5.0
+beartype==0.19.0
+propcache==0.3.1
+typeguard==4.3.0
+torchaudio==2.7.1+cu128
+decorator==4.4.2
+nvidia-cudnn-cu12==9.7.1.26
+torchvision==0.22.1+cu128
+sentry-sdk==2.25.1
+mdurl==0.1.2
+more-itertools==10.3.0
+einops==0.8.1
+zipp==3.19.2
+antlr4-python3-runtime==4.9.3
+cycler==0.12.1
+nvidia-cufile-cu12==1.13.0.11
+jaraco.collections==5.1.0
+jaraco.text==3.12.1
+torchmetrics==1.8.2
+packaging==24.2
+black==25.1.0
+setproctitle==1.3.5
+Pygments==2.19.2
+tomli==2.2.1
+six==1.17.0
+docker-pycreds==0.4.0
+inflect==7.3.1
+requests==2.28.1
+svg.py==1.6.0
+python-dateutil==2.9.0.post0
+proglog==0.1.11
+gsplat==1.5.3
+e3nn==0.5.6
+pathspec==0.12.1
+aiosignal==1.3.2
+wheel==0.45.1
+multidict==6.3.2
+pyparsing==3.2.3
+plyfile==1.1
+jaraco.functools==4.0.1
+nvidia-nccl-cu12==2.26.2
+triton==3.3.1
+nvidia-cusparselt-cu12==0.6.3
+scikit-video==1.1.11
+huggingface-hub==0.30.1
+lightning-utilities==0.14.3
+wadler_lindig==0.1.4
+nvidia-curand-cu12==10.3.9.55
+matplotlib==3.10.1
+kiwisolver==1.4.8
+tqdm==4.67.1
+safetensors==0.5.3
+scipy==1.15.2
+lpips==0.1.4
+nvidia-cusolver-cu12==11.7.2.55
+click==8.1.8
+charset-normalizer==2.1.1
+platformdirs==4.3.7
+opt-einsum-fx==0.1.4
+annotated-types==0.7.0
+smmap==5.0.2
+gitdb==4.0.12
+nvidia-cufft-cu12==11.3.3.41
+contourpy==1.3.1
+urllib3==1.26.13
+autocommand==2.2.2
+yarl==1.19.0
+aiohttp==3.11.16
+colorspacious==1.1.2
+python-dotenv==1.1.0

wandb/run-20250916_180413-d4pobtwb/files/wandb-metadata.json

@@ -0,0 +1,79 @@
+{
+  "os": "Linux-5.15.0-153-generic-x86_64-with-glibc2.35",
+  "python": "CPython 3.12.11",
+  "startedAt": "2025-09-16T10:04:13.466109Z",
+  "args": [
+    "+experiment=re10k",
+    "wandb.mode=online",
+    "wandb.name=re10k_hpc"
+  ],
+  "program": "-m src.main",
+  "git": {
+    "remote": "git@github.com:K-nowing/ICLR2026_HCP.git",
+    "commit": "6c0080eae0295d43dd7f18fd71e742d924238f89"
+  },
+  "email": "dna9041@korea.ac.kr",
+  "root": "/root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc",
+  "host": "notebook-72e474dd-92e1-11f0-a6ea-7ab79f1c53cb-0",
+  "executable": "/opt/conda/envs/ours/bin/python",
+  "cpu_count": 144,
+  "cpu_count_logical": 288,
+  "gpu": "NVIDIA B200",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "1081101176832",
+      "used": "211821613056"
+    }
+  },
+  "memory": {
+    "total": "2434690400256"
+  },
+  "cpu": {
+    "count": 144,
+    "countLogical": 288
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    }
+  ],
+  "cudaVersion": "12.8"
+}

wandb/run-20250916_180413-d4pobtwb/logs/debug-core.log

@@ -0,0 +1,7 @@
+{"time":"2025-09-16T18:04:13.103986093+08:00","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpv2hrevc1/port-6788.txt","pid":6788,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false}
+{"time":"2025-09-16T18:04:13.107503769+08:00","level":"INFO","msg":"Will exit if parent process dies.","ppid":6788}
+{"time":"2025-09-16T18:04:13.107294572+08:00","level":"INFO","msg":"server is running","addr":{"IP":"127.0.0.1","Port":37499,"Zone":""}}
+{"time":"2025-09-16T18:04:13.18584671+08:00","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"127.0.0.1:42398"}
+{"time":"2025-09-16T18:04:13.469622805+08:00","level":"INFO","msg":"handleInformInit: received","streamId":"d4pobtwb","id":"127.0.0.1:42398"}
+{"time":"2025-09-16T18:04:13.944559071+08:00","level":"INFO","msg":"handleInformInit: stream started","streamId":"d4pobtwb","id":"127.0.0.1:42398"}
+{"time":"2025-09-16T18:04:27.095445631+08:00","level":"INFO","msg":"Parent process exited, terminating service process."}

wandb/run-20250916_180413-d4pobtwb/logs/debug-internal.log

@@ -0,0 +1,7 @@
+{"time":"2025-09-16T18:04:13.471240905+08:00","level":"INFO","msg":"stream: starting","core version":"0.19.9","symlink path":"/root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc/wandb/run-20250916_180413-d4pobtwb/logs/debug-core.log"}
+{"time":"2025-09-16T18:04:13.944337794+08:00","level":"INFO","msg":"created new stream","id":"d4pobtwb"}
+{"time":"2025-09-16T18:04:13.944517672+08:00","level":"INFO","msg":"stream: started","id":"d4pobtwb"}
+{"time":"2025-09-16T18:04:13.944738149+08:00","level":"INFO","msg":"writer: Do: started","stream_id":"d4pobtwb"}
+{"time":"2025-09-16T18:04:13.944986426+08:00","level":"INFO","msg":"sender: started","stream_id":"d4pobtwb"}
+{"time":"2025-09-16T18:04:13.945237383+08:00","level":"INFO","msg":"handler: started","stream_id":"d4pobtwb"}
+{"time":"2025-09-16T18:04:14.217447222+08:00","level":"INFO","msg":"Starting system monitor"}

wandb/run-20250916_180413-d4pobtwb/logs/debug.log

@@ -0,0 +1,22 @@
+2025-09-16 18:04:13,440 INFO    MainThread:6788 [wandb_setup.py:_flush():67] Current SDK version is 0.19.9
+2025-09-16 18:04:13,440 INFO    MainThread:6788 [wandb_setup.py:_flush():67] Configure stats pid to 6788
+2025-09-16 18:04:13,440 INFO    MainThread:6788 [wandb_setup.py:_flush():67] Loading settings from /root/.config/wandb/settings
+2025-09-16 18:04:13,440 INFO    MainThread:6788 [wandb_setup.py:_flush():67] Loading settings from /root/highspeedstorage/injae/code/ICLR2026_HCP/wandb/settings
+2025-09-16 18:04:13,440 INFO    MainThread:6788 [wandb_setup.py:_flush():67] Loading settings from environment variables
+2025-09-16 18:04:13,441 INFO    MainThread:6788 [wandb_init.py:setup_run_log_directory():662] Logging user logs to /root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc/wandb/run-20250916_180413-d4pobtwb/logs/debug.log
+2025-09-16 18:04:13,441 INFO    MainThread:6788 [wandb_init.py:setup_run_log_directory():663] Logging internal logs to /root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc/wandb/run-20250916_180413-d4pobtwb/logs/debug-internal.log
+2025-09-16 18:04:13,441 INFO    MainThread:6788 [wandb_init.py:init():781] calling init triggers
+2025-09-16 18:04:13,442 INFO    MainThread:6788 [wandb_init.py:init():786] wandb.init called with sweep_config: {}
+config: {'model': {'encoder': {'backbone': {'name': 'croco', 'model': 'ViTLarge_BaseDecoder', 'patch_embed_cls': 'PatchEmbedDust3R', 'asymmetry_decoder': True, 'intrinsics_embed_loc': 'encoder', 'intrinsics_embed_degree': 4, 'intrinsics_embed_type': 'token'}, 'name': 'noposplat', 'opacity_mapping': {'initial': 0.0, 'final': 0.0, 'warm_up': 1}, 'num_monocular_samples': 32, 'num_surfaces': 1, 'predict_opacity': False, 'gaussians_per_pixel': 1, 'gaussian_adapter': {'gaussian_scale_min': 0.5, 'gaussian_scale_max': 15.0, 'sh_degree': 4}, 'd_feature': 128, 'visualizer': {'num_samples': 8, 'min_resolution': 256, 'export_ply': False}, 'apply_bounds_shim': True, 'gs_params_head_type': 'dpt_gs', 'pose_free': True, 'pretrained_weights': ''}, 'decoder': {'name': 'splatting_cuda', 'background_color': [0.0, 0.0, 0.0], 'make_scale_invariant': False, 'use_gsplat': True}, 'vggt': {'input_image_shape': [518, 518], 'head_mode': 'pcd', 'same_head': False, 'camera_prompt': False, 'num_visual_prompt': 0, 'use_extrinsic_token': False, 'use_intrinsic_token': False, 'detr': {'name': 'deformable_detr', 'features_num_channel': [128, 32], 'hidden_dim': 32, 'dim_down_factor': 1, 'dim_feedforward_factor': 4, 'enc_layers': 6, 'dec_layers': 1, 'dropout': 0.1, 'num_feature_levels': 2, 'dec_n_points': 4, 'enc_n_points': 4, 'split_threshold': 1.0, 'num_queries': 256, 'base_HW': 64, 'position_embedding': 'sine', 'pre_norm': False, 'self_layer_num': 1, 'num_level': 3, 'debug_mode': 0, 'decoder_type': 'conv', 'split_threshold_range': [0.0, 2.0], 'only_last_level': False, 'grad_mode': 'absgrad', 'same_split_head': False, 'use_mean_features': True}}}, 'loss': {'mse': {'weight': 1.0}, 'lpips': {'weight': 0.05, 'apply_after_step': 0}}, 'wandb': {'project': 'noposplat', 'entity': 'scene-representation-group', 'name': 're10k_hpc', 'mode': 'online', 'tags': ['re10k', '256x256']}, 'mode': 'train', 'data_loader': {'train': {'num_workers': 16, 'persistent_workers': True, 'batch_size': 16, 'seed': 1234}, 'test': {'num_workers': 4, 'persistent_workers': False, 'batch_size': 1, 'seed': 2345}, 'val': {'num_workers': 1, 'persistent_workers': True, 'batch_size': 1, 'seed': 3456}}, 'optimizer': {'lr': 0.0002, 'warm_up_steps': 125, 'backbone_lr_multiplier': 0.1, 'backbone_trainable': 'GB+H', 'accumulate': 1}, 'checkpointing': {'load': None, 'every_n_train_steps': 9375, 'save_top_k': 1, 'save_weights_only': True}, 'train': {'depth_mode': None, 'extended_visualization': False, 'print_log_every_n_steps': 10, 'distiller': '', 'distill_max_steps': 1000000, 'debug_mode': 0, 'camera_consis_loss': 0.0, 'camera_loss': 1.0, 'aux_loss': 0.0, 'one_sample_train': False, 'target_scale_mode': 2, 'extrinsic_drop': 0.1, 'intrinsic_drop': 0.1, 'align_corners': False, 'backbone_checkpoint': 'none', 'intrinsic_scaling': True, 'return_level_gs': True, 'train_context': False, 'use_pred_point_focal': False, 'scaling_detach': True, 'split_score_log_scale': True, 'split_score_grad_scale': 10000.0, 'split_loss': 0.01, 'random_split_strategy': 'score'}, 'test': {'output_path': 'test', 'align_pose': False, 'pose_align_steps': 100, 'rot_opt_lr': 0.005, 'trans_opt_lr': 0.005, 'compute_scores': True, 'save_image': False, 'save_video': False, 'save_compare': False, 'split_threshold_list': [0.0, 0.1, 0.2, 0.3, 0.4], 'pred_intrinsic': False}, 'seed': 111123, 'trainer': {'max_steps': 18751, 'val_check_interval': 500, 'gradient_clip_val': 0.5, 'num_nodes': 1}, 'dataset': {'re10k': {'make_baseline_1': True, 'relative_pose': True, 'augment': True, 'background_color': [0.0, 0.0, 0.0], 'overfit_to_scene': None, 'skip_bad_shape': True, 'view_sampler': {'name': 'bounded', 'num_target_views': 4, 'num_context_views': 2, 'min_distance_between_context_views': 45, 'max_distance_between_context_views': 90, 'min_distance_to_context_views': 0, 'warm_up_steps': 9375, 'initial_min_distance_between_context_views': 25, 'initial_max_distance_between_context_views': 25, 'distill_steps': 0}, 'name': 're10k', 'roots': ['datasets/re10k'], 'input_image_shape': [256, 256], 'original_image_shape': [360, 640], 'cameras_are_circular': False, 'baseline_min': 0.001, 'baseline_max': 10000000000.0, 'max_fov': 100.0}}, '_wandb': {}}
+2025-09-16 18:04:13,442 INFO    MainThread:6788 [wandb_init.py:init():809] starting backend
+2025-09-16 18:04:13,442 INFO    MainThread:6788 [wandb_init.py:init():813] sending inform_init request
+2025-09-16 18:04:13,464 INFO    MainThread:6788 [backend.py:_multiprocessing_setup():101] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2025-09-16 18:04:13,464 INFO    MainThread:6788 [wandb_init.py:init():823] backend started and connected
+2025-09-16 18:04:13,488 INFO    MainThread:6788 [wandb_init.py:init():915] updated telemetry
+2025-09-16 18:04:13,535 INFO    MainThread:6788 [wandb_init.py:init():939] communicating run to backend with 90.0 second timeout
+2025-09-16 18:04:14,206 INFO    MainThread:6788 [wandb_init.py:init():1014] starting run threads in backend
+2025-09-16 18:04:14,384 INFO    MainThread:6788 [wandb_run.py:_console_start():2454] atexit reg
+2025-09-16 18:04:14,385 INFO    MainThread:6788 [wandb_run.py:_redirect():2306] redirect: wrap_raw
+2025-09-16 18:04:14,388 INFO    MainThread:6788 [wandb_run.py:_redirect():2371] Wrapping output streams.
+2025-09-16 18:04:14,388 INFO    MainThread:6788 [wandb_run.py:_redirect():2394] Redirects installed.
+2025-09-16 18:04:14,392 INFO    MainThread:6788 [wandb_init.py:init():1056] run started, returning control to user process

wandb/run-20250916_194552-aaciqyuf/files/config.yaml

@@ -0,0 +1,555 @@
+_wandb:
+    value:
+        cli_version: 0.19.9
+        m:
+            - "1": loss/split_score
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": trainer/global_step
+              "6":
+                - 3
+              "7": []
+            - "1": comparison.height
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.captions
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.format
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.count
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.format
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": lr-AdamW/pg2
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": loss/aux_1/mse
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": loss/total
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": val/ssim
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization._type
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.filenames
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": loss/aux_0/mse
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison.captions
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.height
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.count
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": loss/camera
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.filenames
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.width
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": lr-AdamW/pg2-momentum
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": val/lpips
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison.format
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison._type
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.width
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.captions
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": info/global_step
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": loss/aux_2/lpips
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": epoch
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs._type
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.captions
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map._type
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.format
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": lr-AdamW/pg1-momentum
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": loss/aux_1/lpips
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": train/psnr_probabilistic
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": loss/aux_0/lpips
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison.width
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison.count
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": comparison.filenames
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.filenames
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.width
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": activation_visualization.count
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": lr-AdamW/pg1
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": val/gaussian_num_ratio
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": val/psnr
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": points_context_imgs.height
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": attention_map.height
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": loss/3dgs_2_0/mse
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": loss/aux_2/mse
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": loss/3dgs_2_0/lpips
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+        python_version: 3.12.11
+        t:
+            "1":
+                - 1
+                - 41
+                - 49
+                - 50
+                - 55
+                - 106
+            "2":
+                - 1
+                - 41
+                - 49
+                - 50
+                - 55
+                - 106
+            "3":
+                - 7
+                - 13
+                - 15
+                - 16
+                - 23
+                - 55
+                - 66
+            "4": 3.12.11
+            "5": 0.19.9
+            "8":
+                - 5
+            "12": 0.19.9
+            "13": linux-x86_64
+checkpointing:
+    value:
+        every_n_train_steps: 9375
+        load: null
+        save_top_k: 1
+        save_weights_only: true
+data_loader:
+    value:
+        test:
+            batch_size: 1
+            num_workers: 4
+            persistent_workers: false
+            seed: 2345
+        train:
+            batch_size: 16
+            num_workers: 16
+            persistent_workers: true
+            seed: 1234
+        val:
+            batch_size: 1
+            num_workers: 1
+            persistent_workers: true
+            seed: 3456
+dataset:
+    value:
+        re10k:
+            augment: true
+            background_color:
+                - 0
+                - 0
+                - 0
+            baseline_max: 1e+10
+            baseline_min: 0.001
+            cameras_are_circular: false
+            input_image_shape:
+                - 256
+                - 256
+            make_baseline_1: true
+            max_fov: 100
+            name: re10k
+            original_image_shape:
+                - 360
+                - 640
+            overfit_to_scene: null
+            relative_pose: true
+            roots:
+                - datasets/re10k
+            skip_bad_shape: true
+            view_sampler:
+                distill_steps: 0
+                initial_max_distance_between_context_views: 25
+                initial_min_distance_between_context_views: 25
+                max_distance_between_context_views: 90
+                min_distance_between_context_views: 45
+                min_distance_to_context_views: 0
+                name: bounded
+                num_context_views: 2
+                num_target_views: 4
+                warm_up_steps: 9375
+loss:
+    value:
+        lpips:
+            apply_after_step: 0
+            weight: 0.05
+        mse:
+            weight: 1
+mode:
+    value: train
+model:
+    value:
+        decoder:
+            background_color:
+                - 0
+                - 0
+                - 0
+            make_scale_invariant: false
+            name: splatting_cuda
+            use_gsplat: true
+        encoder:
+            apply_bounds_shim: true
+            backbone:
+                asymmetry_decoder: true
+                intrinsics_embed_degree: 4
+                intrinsics_embed_loc: encoder
+                intrinsics_embed_type: token
+                model: ViTLarge_BaseDecoder
+                name: croco
+                patch_embed_cls: PatchEmbedDust3R
+            d_feature: 128
+            gaussian_adapter:
+                gaussian_scale_max: 15
+                gaussian_scale_min: 0.5
+                sh_degree: 4
+            gaussians_per_pixel: 1
+            gs_params_head_type: dpt_gs
+            name: noposplat
+            num_monocular_samples: 32
+            num_surfaces: 1
+            opacity_mapping:
+                final: 0
+                initial: 0
+                warm_up: 1
+            pose_free: true
+            predict_opacity: false
+            pretrained_weights: ""
+            visualizer:
+                export_ply: false
+                min_resolution: 256
+                num_samples: 8
+        vggt:
+            camera_prompt: false
+            detr:
+                base_HW: 64
+                debug_mode: 0
+                dec_layers: 1
+                dec_n_points: 4
+                decoder_type: conv
+                dim_down_factor: 1
+                dim_feedforward_factor: 4
+                dropout: 0.1
+                enc_layers: 6
+                enc_n_points: 4
+                features_num_channel:
+                    - 128
+                    - 32
+                grad_mode: absgrad
+                hidden_dim: 32
+                name: deformable_detr
+                num_feature_levels: 2
+                num_level: 3
+                num_queries: 256
+                only_last_level: false
+                position_embedding: sine
+                pre_norm: false
+                same_split_head: false
+                self_layer_num: 1
+                split_threshold: 1
+                split_threshold_range:
+                    - 0
+                    - 2
+                use_mean_features: true
+            head_mode: pcd
+            input_image_shape:
+                - 518
+                - 518
+            num_visual_prompt: 0
+            same_head: false
+            use_extrinsic_token: false
+            use_intrinsic_token: false
+optimizer:
+    value:
+        accumulate: 1
+        backbone_lr_multiplier: 0.1
+        backbone_trainable: GB+H
+        lr: 0.0002
+        warm_up_steps: 125
+seed:
+    value: 111123
+test:
+    value:
+        align_pose: false
+        compute_scores: true
+        output_path: test
+        pose_align_steps: 100
+        pred_intrinsic: false
+        rot_opt_lr: 0.005
+        save_compare: false
+        save_image: false
+        save_video: false
+        split_threshold_list:
+            - 0
+            - 0.1
+            - 0.2
+            - 0.3
+            - 0.4
+        trans_opt_lr: 0.005
+train:
+    value:
+        align_corners: false
+        aux_loss: 0
+        backbone_checkpoint: none
+        camera_consis_loss: 0
+        camera_loss: 1
+        debug_mode: 0
+        depth_mode: null
+        distill_max_steps: 1000000
+        distiller: ""
+        extended_visualization: false
+        extrinsic_drop: 0.1
+        intrinsic_drop: 0.1
+        intrinsic_scaling: true
+        one_sample_train: false
+        print_log_every_n_steps: 10
+        random_split_strategy: score
+        return_level_gs: true
+        scaling_detach: true
+        split_loss: 0.01
+        split_score_grad_scale: 10000
+        split_score_log_scale: true
+        target_scale_mode: 2
+        train_context: false
+        use_pred_point_focal: false
+trainer:
+    value:
+        gradient_clip_val: 0.5
+        max_steps: 18751
+        num_nodes: 1
+        val_check_interval: 500
+wandb:
+    value:
+        entity: scene-representation-group
+        mode: online
+        name: re10k_hpc
+        project: noposplat
+        tags:
+            - re10k
+            - 256x256

wandb/run-20250916_194552-aaciqyuf/files/requirements.txt

@@ -0,0 +1,129 @@
+ruff==0.11.4
+PyYAML==6.0.2
+sympy==1.13.3
+fsspec==2024.6.1
+imageio-ffmpeg==0.6.0
+setuptools==78.1.1
+timm==1.0.15
+fonttools==4.57.0
+nvidia-cublas-cu12==12.8.3.14
+nvidia-cuda-nvrtc-cu12==12.8.61
+dacite==1.9.2
+hydra-core==1.3.2
+certifi==2022.12.7
+rich==14.0.0
+filelock==3.13.1
+idna==3.4
+nvidia-cuda-cupti-cu12==12.8.57
+pip==25.2
+wandb==0.19.9
+ninja==1.11.1.4
+networkx==3.3
+GitPython==3.1.44
+moviepy==1.0.3
+markdown-it-py==3.0.0
+lightning==2.5.1
+Jinja2==3.1.4
+nvidia-cusparse-cu12==12.5.7.53
+tabulate==0.9.0
+tifffile==2025.3.30
+nvidia-nvtx-cu12==12.8.55
+jaxtyping==0.2.37
+aiohappyeyeballs==2.6.1
+MarkupSafe==2.1.5
+pydantic_core==2.33.1
+mypy-extensions==1.0.0
+lazy_loader==0.4
+pydantic==2.11.2
+typing-inspection==0.4.0
+pillow==10.4.0
+nvidia-nvjitlink-cu12==12.8.61
+omegaconf==2.3.0
+scikit-image==0.25.2
+jaraco.context==5.3.0
+typing_extensions==4.12.2
+mpmath==1.3.0
+opencv-python==4.11.0.86
+backports.tarfile==1.2.0
+async-timeout==5.0.1
+opt_einsum==3.4.0
+nvidia-cuda-runtime-cu12==12.8.57
+numpy==1.26.4
+torch==2.7.1+cu128
+psutil==7.0.0
+pytorch-lightning==2.5.1
+protobuf==5.29.4
+imageio==2.37.0
+colorama==0.4.6
+attrs==25.3.0
+importlib_metadata==8.0.0
+frozenlist==1.5.0
+beartype==0.19.0
+propcache==0.3.1
+typeguard==4.3.0
+torchaudio==2.7.1+cu128
+decorator==4.4.2
+nvidia-cudnn-cu12==9.7.1.26
+torchvision==0.22.1+cu128
+sentry-sdk==2.25.1
+mdurl==0.1.2
+more-itertools==10.3.0
+einops==0.8.1
+zipp==3.19.2
+antlr4-python3-runtime==4.9.3
+cycler==0.12.1
+nvidia-cufile-cu12==1.13.0.11
+jaraco.collections==5.1.0
+jaraco.text==3.12.1
+torchmetrics==1.8.2
+packaging==24.2
+black==25.1.0
+setproctitle==1.3.5
+Pygments==2.19.2
+tomli==2.2.1
+six==1.17.0
+docker-pycreds==0.4.0
+inflect==7.3.1
+requests==2.28.1
+svg.py==1.6.0
+python-dateutil==2.9.0.post0
+proglog==0.1.11
+gsplat==1.5.3
+e3nn==0.5.6
+pathspec==0.12.1
+aiosignal==1.3.2
+wheel==0.45.1
+multidict==6.3.2
+pyparsing==3.2.3
+plyfile==1.1
+jaraco.functools==4.0.1
+nvidia-nccl-cu12==2.26.2
+triton==3.3.1
+nvidia-cusparselt-cu12==0.6.3
+scikit-video==1.1.11
+huggingface-hub==0.30.1
+lightning-utilities==0.14.3
+wadler_lindig==0.1.4
+nvidia-curand-cu12==10.3.9.55
+matplotlib==3.10.1
+kiwisolver==1.4.8
+tqdm==4.67.1
+safetensors==0.5.3
+scipy==1.15.2
+lpips==0.1.4
+nvidia-cusolver-cu12==11.7.2.55
+click==8.1.8
+charset-normalizer==2.1.1
+platformdirs==4.3.7
+opt-einsum-fx==0.1.4
+annotated-types==0.7.0
+smmap==5.0.2
+gitdb==4.0.12
+nvidia-cufft-cu12==11.3.3.41
+contourpy==1.3.1
+urllib3==1.26.13
+autocommand==2.2.2
+yarl==1.19.0
+aiohttp==3.11.16
+colorspacious==1.1.2
+python-dotenv==1.1.0

wandb/run-20250916_194552-aaciqyuf/files/wandb-metadata.json

@@ -0,0 +1,79 @@
+{
+  "os": "Linux-5.15.0-153-generic-x86_64-with-glibc2.35",
+  "python": "CPython 3.12.11",
+  "startedAt": "2025-09-16T11:45:52.656293Z",
+  "args": [
+    "+experiment=re10k",
+    "wandb.mode=online",
+    "wandb.name=re10k_hpc"
+  ],
+  "program": "-m src.main",
+  "git": {
+    "remote": "git@github.com:K-nowing/ICLR2026_HCP.git",
+    "commit": "6c0080eae0295d43dd7f18fd71e742d924238f89"
+  },
+  "email": "dna9041@korea.ac.kr",
+  "root": "/root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc",
+  "host": "notebook-72e474dd-92e1-11f0-a6ea-7ab79f1c53cb-0",
+  "executable": "/opt/conda/envs/ours/bin/python",
+  "cpu_count": 144,
+  "cpu_count_logical": 288,
+  "gpu": "NVIDIA B200",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "1081101176832",
+      "used": "208657956864"
+    }
+  },
+  "memory": {
+    "total": "2434690371584"
+  },
+  "cpu": {
+    "count": 144,
+    "countLogical": 288
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    },
+    {
+      "name": "NVIDIA B200",
+      "memoryTotal": "192265846784",
+      "cudaCores": 18944
+    }
+  ],
+  "cudaVersion": "12.8"
+}

wandb/run-20250916_194552-aaciqyuf/files/wandb-summary.json

@@ -0,0 +1 @@
+{"lr-AdamW/pg2":2e-05,"loss/aux_0/mse":0.0034407840576022863,"attention_map":{"count":1,"filenames":["media/images/attention_map_277_0b32a0de1e86a80b9b35.png"],"captions":["651a7f83ed093001"],"_type":"images/separated","width":800,"height":536,"format":"png"},"val/gaussian_num_ratio":0.5851516723632812,"loss/total":0.03648084029555321,"lr-AdamW/pg2-momentum":0.9,"loss/aux_0/lpips":0.007770626340061426,"loss/aux_2/mse":0.0024059175048023462,"loss/aux_1/lpips":0.005286840721964836,"val/psnr":27.459697723388672,"_wandb":{"runtime":15987},"trainer/global_step":5599,"loss/aux_2/lpips":0.004525118973106146,"lr-AdamW/pg1-momentum":0.9,"loss/split_score":0.002340342616662383,"loss/3dgs_2_0/lpips":0.005342466291040182,"points_context_imgs":{"format":"png","count":1,"filenames":["media/images/points_context_imgs_275_a06eaa47ddd5bc5fb373.png"],"captions":["651a7f83ed093001"],"_type":"images/separated","width":536,"height":800},"activation_visualization":{"width":256,"height":256,"format":"png","count":1,"filenames":["media/images/activation_visualization_276_cc0deded106bd8c609c8.png"],"captions":["651a7f83ed093001"],"_type":"images/separated"},"_timestamp":1.7580390648492198e+09,"loss/aux_1/mse":0.0026043567340821028,"val/ssim":0.8060880303382874,"loss/camera":0.00019968389824498445,"epoch":0,"train/psnr_probabilistic":26.956830978393555,"val/lpips":0.12397965788841248,"info/global_step":5600,"comparison":{"captions":["651a7f83ed093001"],"_type":"images/separated","width":1064,"height":1098,"format":"png","count":1,"filenames":["media/images/comparison_274_cbe0488a19d085054063.png"]},"_step":282,"lr-AdamW/pg1":0.00016472919392187862,"_runtime":15912.194091688,"loss/3dgs_2_0/mse":0.0025647059082984924}

wandb/run-20250916_194552-aaciqyuf/logs/debug-core.log

@@ -0,0 +1,13 @@
+{"time":"2025-09-16T19:45:52.307943975+08:00","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpg2d9js4m/port-217.txt","pid":217,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false}
+{"time":"2025-09-16T19:45:52.319896951+08:00","level":"INFO","msg":"Will exit if parent process dies.","ppid":217}
+{"time":"2025-09-16T19:45:52.326518926+08:00","level":"INFO","msg":"server is running","addr":{"IP":"127.0.0.1","Port":37777,"Zone":""}}
+{"time":"2025-09-16T19:45:52.440206332+08:00","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"127.0.0.1:46444"}
+{"time":"2025-09-16T19:45:52.665531521+08:00","level":"INFO","msg":"handleInformInit: received","streamId":"aaciqyuf","id":"127.0.0.1:46444"}
+{"time":"2025-09-16T19:45:54.002875103+08:00","level":"INFO","msg":"handleInformInit: stream started","streamId":"aaciqyuf","id":"127.0.0.1:46444"}
+{"time":"2025-09-17T00:12:20.364469246+08:00","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"127.0.0.1:46444"}
+{"time":"2025-09-17T00:12:20.364629241+08:00","level":"INFO","msg":"connection: closing","id":"127.0.0.1:46444"}
+{"time":"2025-09-17T00:12:20.364808859+08:00","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-09-17T00:12:20.364931191+08:00","level":"INFO","msg":"connection: closed successfully","id":"127.0.0.1:46444"}
+{"time":"2025-09-17T00:12:20.884683539+08:00","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"127.0.0.1:46444"}
+{"time":"2025-09-17T00:12:20.884731379+08:00","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"127.0.0.1:46444"}
+{"time":"2025-09-17T00:12:20.884751901+08:00","level":"INFO","msg":"server is closed"}

wandb/run-20250916_194552-aaciqyuf/logs/debug-internal.log

@@ -0,0 +1,15 @@
+{"time":"2025-09-16T19:45:52.670660357+08:00","level":"INFO","msg":"stream: starting","core version":"0.19.9","symlink path":"/root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc/wandb/run-20250916_194552-aaciqyuf/logs/debug-core.log"}
+{"time":"2025-09-16T19:45:54.002750276+08:00","level":"INFO","msg":"created new stream","id":"aaciqyuf"}
+{"time":"2025-09-16T19:45:54.002860648+08:00","level":"INFO","msg":"stream: started","id":"aaciqyuf"}
+{"time":"2025-09-16T19:45:54.003038437+08:00","level":"INFO","msg":"writer: Do: started","stream_id":"aaciqyuf"}
+{"time":"2025-09-16T19:45:54.003259991+08:00","level":"INFO","msg":"sender: started","stream_id":"aaciqyuf"}
+{"time":"2025-09-16T19:45:54.003399136+08:00","level":"INFO","msg":"handler: started","stream_id":"aaciqyuf"}
+{"time":"2025-09-16T19:45:54.411678515+08:00","level":"INFO","msg":"Starting system monitor"}
+{"time":"2025-09-17T00:12:20.36477451+08:00","level":"INFO","msg":"stream: closing","id":"aaciqyuf"}
+{"time":"2025-09-17T00:12:20.364855166+08:00","level":"INFO","msg":"Stopping system monitor"}
+{"time":"2025-09-17T00:12:20.36503602+08:00","level":"INFO","msg":"Stopped system monitor"}
+{"time":"2025-09-17T00:12:20.743361444+08:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-09-17T00:12:20.884041581+08:00","level":"INFO","msg":"handler: closed","stream_id":"aaciqyuf"}
+{"time":"2025-09-17T00:12:20.884178301+08:00","level":"INFO","msg":"sender: closed","stream_id":"aaciqyuf"}
+{"time":"2025-09-17T00:12:20.884170005+08:00","level":"INFO","msg":"writer: Close: closed","stream_id":"aaciqyuf"}
+{"time":"2025-09-17T00:12:20.884479503+08:00","level":"INFO","msg":"stream: closed","id":"aaciqyuf"}

wandb/run-20250916_194552-aaciqyuf/logs/debug.log

@@ -0,0 +1,364 @@
+2025-09-16 19:45:52,629 INFO    MainThread:217 [wandb_setup.py:_flush():67] Current SDK version is 0.19.9
+2025-09-16 19:45:52,629 INFO    MainThread:217 [wandb_setup.py:_flush():67] Configure stats pid to 217
+2025-09-16 19:45:52,629 INFO    MainThread:217 [wandb_setup.py:_flush():67] Loading settings from /root/.config/wandb/settings
+2025-09-16 19:45:52,629 INFO    MainThread:217 [wandb_setup.py:_flush():67] Loading settings from /root/highspeedstorage/injae/code/ICLR2026_HCP/wandb/settings
+2025-09-16 19:45:52,629 INFO    MainThread:217 [wandb_setup.py:_flush():67] Loading settings from environment variables
+2025-09-16 19:45:52,630 INFO    MainThread:217 [wandb_init.py:setup_run_log_directory():662] Logging user logs to /root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc/wandb/run-20250916_194552-aaciqyuf/logs/debug.log
+2025-09-16 19:45:52,630 INFO    MainThread:217 [wandb_init.py:setup_run_log_directory():663] Logging internal logs to /root/highspeedstorage/injae/code/ICLR2026_HCP/outputs/re10k_hpc/wandb/run-20250916_194552-aaciqyuf/logs/debug-internal.log
+2025-09-16 19:45:52,630 INFO    MainThread:217 [wandb_init.py:init():781] calling init triggers
+2025-09-16 19:45:52,631 INFO    MainThread:217 [wandb_init.py:init():786] wandb.init called with sweep_config: {}
+config: {'model': {'encoder': {'backbone': {'name': 'croco', 'model': 'ViTLarge_BaseDecoder', 'patch_embed_cls': 'PatchEmbedDust3R', 'asymmetry_decoder': True, 'intrinsics_embed_loc': 'encoder', 'intrinsics_embed_degree': 4, 'intrinsics_embed_type': 'token'}, 'name': 'noposplat', 'opacity_mapping': {'initial': 0.0, 'final': 0.0, 'warm_up': 1}, 'num_monocular_samples': 32, 'num_surfaces': 1, 'predict_opacity': False, 'gaussians_per_pixel': 1, 'gaussian_adapter': {'gaussian_scale_min': 0.5, 'gaussian_scale_max': 15.0, 'sh_degree': 4}, 'd_feature': 128, 'visualizer': {'num_samples': 8, 'min_resolution': 256, 'export_ply': False}, 'apply_bounds_shim': True, 'gs_params_head_type': 'dpt_gs', 'pose_free': True, 'pretrained_weights': ''}, 'decoder': {'name': 'splatting_cuda', 'background_color': [0.0, 0.0, 0.0], 'make_scale_invariant': False, 'use_gsplat': True}, 'vggt': {'input_image_shape': [518, 518], 'head_mode': 'pcd', 'same_head': False, 'camera_prompt': False, 'num_visual_prompt': 0, 'use_extrinsic_token': False, 'use_intrinsic_token': False, 'detr': {'name': 'deformable_detr', 'features_num_channel': [128, 32], 'hidden_dim': 32, 'dim_down_factor': 1, 'dim_feedforward_factor': 4, 'enc_layers': 6, 'dec_layers': 1, 'dropout': 0.1, 'num_feature_levels': 2, 'dec_n_points': 4, 'enc_n_points': 4, 'split_threshold': 1.0, 'num_queries': 256, 'base_HW': 64, 'position_embedding': 'sine', 'pre_norm': False, 'self_layer_num': 1, 'num_level': 3, 'debug_mode': 0, 'decoder_type': 'conv', 'split_threshold_range': [0.0, 2.0], 'only_last_level': False, 'grad_mode': 'absgrad', 'same_split_head': False, 'use_mean_features': True}}}, 'loss': {'mse': {'weight': 1.0}, 'lpips': {'weight': 0.05, 'apply_after_step': 0}}, 'wandb': {'project': 'noposplat', 'entity': 'scene-representation-group', 'name': 're10k_hpc', 'mode': 'online', 'tags': ['re10k', '256x256']}, 'mode': 'train', 'data_loader': {'train': {'num_workers': 16, 'persistent_workers': True, 'batch_size': 16, 'seed': 1234}, 'test': {'num_workers': 4, 'persistent_workers': False, 'batch_size': 1, 'seed': 2345}, 'val': {'num_workers': 1, 'persistent_workers': True, 'batch_size': 1, 'seed': 3456}}, 'optimizer': {'lr': 0.0002, 'warm_up_steps': 125, 'backbone_lr_multiplier': 0.1, 'backbone_trainable': 'GB+H', 'accumulate': 1}, 'checkpointing': {'load': None, 'every_n_train_steps': 9375, 'save_top_k': 1, 'save_weights_only': True}, 'train': {'depth_mode': None, 'extended_visualization': False, 'print_log_every_n_steps': 10, 'distiller': '', 'distill_max_steps': 1000000, 'debug_mode': 0, 'camera_consis_loss': 0.0, 'camera_loss': 1.0, 'aux_loss': 0.0, 'one_sample_train': False, 'target_scale_mode': 2, 'extrinsic_drop': 0.1, 'intrinsic_drop': 0.1, 'align_corners': False, 'backbone_checkpoint': 'none', 'intrinsic_scaling': True, 'return_level_gs': True, 'train_context': False, 'use_pred_point_focal': False, 'scaling_detach': True, 'split_score_log_scale': True, 'split_score_grad_scale': 10000.0, 'split_loss': 0.01, 'random_split_strategy': 'score'}, 'test': {'output_path': 'test', 'align_pose': False, 'pose_align_steps': 100, 'rot_opt_lr': 0.005, 'trans_opt_lr': 0.005, 'compute_scores': True, 'save_image': False, 'save_video': False, 'save_compare': False, 'split_threshold_list': [0.0, 0.1, 0.2, 0.3, 0.4], 'pred_intrinsic': False}, 'seed': 111123, 'trainer': {'max_steps': 18751, 'val_check_interval': 500, 'gradient_clip_val': 0.5, 'num_nodes': 1}, 'dataset': {'re10k': {'make_baseline_1': True, 'relative_pose': True, 'augment': True, 'background_color': [0.0, 0.0, 0.0], 'overfit_to_scene': None, 'skip_bad_shape': True, 'view_sampler': {'name': 'bounded', 'num_target_views': 4, 'num_context_views': 2, 'min_distance_between_context_views': 45, 'max_distance_between_context_views': 90, 'min_distance_to_context_views': 0, 'warm_up_steps': 9375, 'initial_min_distance_between_context_views': 25, 'initial_max_distance_between_context_views': 25, 'distill_steps': 0}, 'name': 're10k', 'roots': ['datasets/re10k'], 'input_image_shape': [256, 256], 'original_image_shape': [360, 640], 'cameras_are_circular': False, 'baseline_min': 0.001, 'baseline_max': 10000000000.0, 'max_fov': 100.0}}, '_wandb': {}}
+2025-09-16 19:45:52,631 INFO    MainThread:217 [wandb_init.py:init():809] starting backend
+2025-09-16 19:45:52,631 INFO    MainThread:217 [wandb_init.py:init():813] sending inform_init request
+2025-09-16 19:45:52,655 INFO    MainThread:217 [backend.py:_multiprocessing_setup():101] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2025-09-16 19:45:52,655 INFO    MainThread:217 [wandb_init.py:init():823] backend started and connected
+2025-09-16 19:45:52,668 INFO    MainThread:217 [wandb_init.py:init():915] updated telemetry
+2025-09-16 19:45:52,704 INFO    MainThread:217 [wandb_init.py:init():939] communicating run to backend with 90.0 second timeout
+2025-09-16 19:45:54,358 INFO    MainThread:217 [wandb_init.py:init():1014] starting run threads in backend
+2025-09-16 19:45:54,813 INFO    MainThread:217 [wandb_run.py:_console_start():2454] atexit reg
+2025-09-16 19:45:54,814 INFO    MainThread:217 [wandb_run.py:_redirect():2306] redirect: wrap_raw
+2025-09-16 19:45:54,814 INFO    MainThread:217 [wandb_run.py:_redirect():2371] Wrapping output streams.
+2025-09-16 19:45:54,814 INFO    MainThread:217 [wandb_run.py:_redirect():2394] Redirects installed.
+2025-09-16 19:45:54,833 INFO    MainThread:217 [wandb_init.py:init():1056] run started, returning control to user process
+2025-09-17 00:12:20,362 INFO    MsgRouterThr:217 [mailbox.py:close():129] [no run ID] Closing mailbox, abandoning 1 handles.
+2025-09-17 00:12:20,841 ERROR   MainThread:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe
+2025-09-17 00:12:20,882 ERROR   MainThread:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe
+2025-09-17 00:12:20,882 ERROR   MainThread:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe
+2025-09-17 00:12:20,882 ERROR   MainThread:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe
+2025-09-17 00:12:20,883 ERROR   MainThread:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe
+2025-09-17 00:12:20,886 ERROR   MainThread:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe
+2025-09-17 00:12:20,887 ERROR   MainThread:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe
+2025-09-17 00:12:20,887 ERROR   MainThread:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe
+2025-09-17 00:12:20,887 ERROR   MainThread:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe
+2025-09-17 00:12:20,888 ERROR   MainThread:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe
+2025-09-17 00:12:21,715 ERROR   child-process-observer:217 [redirect.py:_on_write():661] [no run ID] error in stderr callback
+Traceback (most recent call last):
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/redirect.py", line 659, in _on_write
+    cb(written_data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 2381, in <lambda>
+    lambda data: self._console_raw_callback("stderr", data),
+                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 387, in wrapper
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 435, in wrapper_fn
+    return func(self, *args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/wandb_run.py", line 1518, in _console_raw_callback
+    self._backend.interface.publish_output_raw(name, data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface.py", line 761, in publish_output_raw
+    self._publish_output_raw(o)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_shared.py", line 38, in _publish_output_raw
+    self._publish(rec)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/interface/interface_sock.py", line 39, in _publish
+    self._sock_client.send_record_publish(record)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 174, in send_record_publish
+    self.send_server_request(server_req)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 154, in send_server_request
+    self._send_message(msg)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 151, in _send_message
+    self._sendall_with_error_handle(header + data)
+  File "/opt/conda/envs/ours/lib/python3.12/site-packages/wandb/sdk/lib/sock_client.py", line 130, in _sendall_with_error_handle
+    sent = self._sock.send(data)
+           ^^^^^^^^^^^^^^^^^^^^^
+BrokenPipeError: [Errno 32] Broken pipe