Datasets:

Knowing
/

2

+model:
+  encoder:
+    name: dcsplat
+    input_image_shape:
+    - 518
+    - 518
+    head_mode: pcd
+    num_level: 3
+    gs_param_dim: 256
+    align_corners: false
+    use_voxelize: true
+  decoder:
+    name: splatting_cuda
+    background_color:
+    - 0.0
+    - 0.0
+    - 0.0
+    make_scale_invariant: false
+  density_control:
+    name: density_control_module
+    mean_dim: 32
+    gs_param_dim: 256
+    refinement_layer_num: 1
+    num_level: 3
+    grad_mode: absgrad
+    use_mean_features: true
+    refinement_type: voxelize
+    refinement_hidden_dim: 32
+    aggregation_mode: mean
+    num_heads: 1
+    score_mode: frequency
+    latent_dim: 128
+    num_latents: 64
+    num_self_attn_per_block: 2
+    voxel_size: 0.001
+    aux_refine: false
+    refine_error: false
+    use_refine_module: false
+    voxelize_activate: false
+    use_depth: false
+render_loss:
+  mse:
+    weight: 1.0
+  lpips:
+    weight: 0.05
+    apply_after_step: 0
+density_control_loss:
+  error_score:
+    weight: 0.0001
+    log_scale: false
+    grad_scale: 10000.0
+    mode: original
+direct_loss:
+  l1:
+    weight: 0.8
+  ssim:
+    weight: 0.2
+wandb:
+  project: DCSplat
+  entity: scene-representation-group
+  name: ABLATION_0302_FreqSelect
+  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: 25
+  backbone_lr_multiplier: 0.1
+  backbone_trainable: T+H
+  accumulate: 1
+checkpointing:
+  load: null
+  every_n_train_steps: 1500
+  save_top_k: 2
+  save_weights_only: false
+train:
+  extended_visualization: false
+  print_log_every_n_steps: 10
+  camera_loss: 10.0
+  one_sample_validation: null
+  align_corners: false
+  intrinsic_scaling: false
+  verbose: false
+  beta_dist_param:
+  - 0.5
+  - 4.0
+  use_refine_aux: false
+  train_target_set: true
+  train_gs_num: 1
+  ext_scale_detach: false
+  cam_scale_mode: sum
+  scene_scale_reg_loss: 0.01
+  train_aux: true
+  vggt_cam_loss: true
+  vggt_distil: false
+  context_view_train: false
+test:
+  output_path: test/ablation/re10k
+  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_active_mask_image: false
+  save_error_score_image: false
+  save_compare: false
+  save_gs: false
+  save_sample_wise_metrics: true
+  pred_intrinsic: false
+  error_threshold: 0.4
+  error_threshold_list:
+  - 0.2
+  - 0.4
+  - 0.6
+  - 0.8
+  - 1.0
+  threshold_mode: ratio
+  nvs_view_N_list:
+  - 3
+  - 6
+  - 16
+  - 32
+  - 64
+seed: 111123
+trainer:
+  max_steps: 3001
+  val_check_interval: 250
+  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: 1000
+      initial_min_distance_between_context_views: 25
+      initial_max_distance_between_context_views: 25
+      same_target_gap: false
+      num_target_set: 3
+      target_align: true
+    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
+    dynamic_context_views: true
+    max_context_views_per_gpu: 24

ABLATION_0302_FreqSelect/.hydra/hydra.yaml ADDED Viewed

	@@ -0,0 +1,165 @@

+hydra:
+  run:
+    dir: outputs/ablation/re10k/${wandb.name}
+  sweep:
+    dir: multirun/${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_ablation_24v
+    - wandb.mode=online
+    - wandb.name=ABLATION_0302_FreqSelect
+    - model.density_control.score_mode=frequency
+  job:
+    name: main
+    chdir: null
+    override_dirname: +experiment=re10k_ablation_24v,model.density_control.score_mode=frequency,wandb.mode=online,wandb.name=ABLATION_0302_FreqSelect
+    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: /workspace/code/CVPR2026
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /workspace/code/CVPR2026/config
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /workspace/code/CVPR2026/outputs/ablation/re10k/ABLATION_0302_FreqSelect
+    choices:
+      experiment: re10k_ablation_24v
+      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/density_control: density_control_module
+      model/decoder: splatting_cuda
+      model/encoder: dcsplat
+      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

ABLATION_0302_FreqSelect/.hydra/overrides.yaml ADDED Viewed

	@@ -0,0 +1,4 @@

+- +experiment=re10k_ablation_24v
+- wandb.mode=online
+- wandb.name=ABLATION_0302_FreqSelect
+- model.density_control.score_mode=frequency

ABLATION_0302_FreqSelect/main.log ADDED Viewed

	@@ -0,0 +1,128 @@

+[2026-03-02 16:44:31,949][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 16:44:38,058][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 16:44:38,058][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 16:45:26,165][py.warnings][WARNING] - /venv/main/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=31` in the `DataLoader` to improve performance.
+[2026-03-02 16:45:26,166][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 16:45:28,601][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 16:45:28,610][py.warnings][WARNING] - /venv/main/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)`.
+[2026-03-02 16:45:28,611][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 16:45:28,611][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 16:45:30,262][py.warnings][WARNING] - /venv/main/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:4322.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+[2026-03-02 16:45:30,551][py.warnings][WARNING] - /venv/main/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.
+[2026-03-02 16:45:30,553][py.warnings][WARNING] - /venv/main/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.
+[2026-03-02 16:45:30,553][py.warnings][WARNING] - /venv/main/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.
+[2026-03-02 16:45:30,553][py.warnings][WARNING] - /venv/main/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.
+[2026-03-02 16:45:30,554][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('info/global_step', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+[2026-03-02 16:45:40,495][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 16:45:40,609][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 16:47:13,608][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-02 16:58:03,116][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:01:13,793][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:10:31,607][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:16:40,882][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:22:56,474][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:32:09,351][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:35:19,468][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:47:34,983][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:47:38,498][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:00:01,397][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:03:05,664][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:12:22,306][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:18:46,094][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:24:57,888][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:34:16,505][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:37:24,973][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:49:40,257][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:49:44,523][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:02:07,064][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:05:16,184][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:14:35,746][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:20:41,402][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:26:53,279][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:36:05,438][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:39:14,878][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:51:47,692][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:51:51,180][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_FreqSelect/train_ddp_process_1.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-02 16:44:48,362][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 16:45:07,205][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 16:45:07,209][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 16:45:26,166][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 16:45:40,490][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 16:45:40,724][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 16:47:13,415][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-02 16:58:03,116][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:10:31,607][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:22:56,474][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:35:19,469][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:47:38,498][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:00:01,397][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:12:22,307][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:24:57,886][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:37:24,971][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:49:44,522][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:02:07,064][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:14:35,747][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:26:53,279][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:39:14,878][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:51:51,180][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_FreqSelect/train_ddp_process_2.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-02 16:44:48,522][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 16:45:16,036][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 16:45:16,036][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 16:45:26,166][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 16:45:40,493][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 16:45:40,600][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 16:47:13,415][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-02 16:58:03,117][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:10:31,607][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:22:56,474][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:35:19,468][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:47:38,499][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:00:01,397][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:12:22,307][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:24:57,886][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:37:24,972][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:49:44,522][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:02:07,064][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:14:35,747][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:26:53,279][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:39:14,878][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:51:51,181][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_FreqSelect/train_ddp_process_3.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-02 16:44:48,521][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 16:45:15,632][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 16:45:15,633][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 16:45:26,166][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 16:45:40,499][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 16:45:40,619][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 16:47:13,444][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-02 16:58:03,117][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:10:31,608][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:22:56,475][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:35:19,468][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:47:38,498][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:00:01,398][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:12:22,307][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:24:57,887][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:37:24,971][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:49:44,521][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:02:07,064][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:14:35,748][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:26:53,280][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:39:14,878][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:51:51,180][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_FreqSelect/train_ddp_process_4.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-02 16:44:48,474][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 16:45:06,776][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 16:45:06,778][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 16:45:26,166][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 16:45:40,492][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 16:45:40,622][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 16:47:13,433][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-02 16:58:03,116][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:10:31,606][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:22:56,474][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:35:19,468][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:47:38,498][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:00:01,397][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:12:22,306][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:24:57,886][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:37:24,971][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:49:44,522][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:02:07,065][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:14:35,747][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:26:53,279][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:39:14,878][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:51:51,180][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_FreqSelect/train_ddp_process_5.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-02 16:44:48,532][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 16:45:15,740][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 16:45:15,744][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 16:45:26,166][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 16:45:39,751][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 16:45:40,609][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 16:47:13,436][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-02 16:58:03,116][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:10:31,606][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:22:56,474][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:35:19,468][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:47:38,498][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:00:01,397][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:12:22,307][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:24:57,886][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:37:24,971][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:49:44,521][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:02:07,064][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:14:35,746][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:26:53,280][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:39:15,134][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:51:51,180][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_FreqSelect/train_ddp_process_6.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-02 16:44:48,530][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 16:45:15,975][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 16:45:15,975][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 16:45:26,166][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 16:45:39,979][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 16:45:40,604][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 16:47:13,416][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-02 16:58:03,116][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:10:31,606][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:22:56,474][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:35:19,468][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:47:38,498][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:00:01,397][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:12:22,307][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:24:57,886][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:37:24,971][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:49:44,521][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:02:07,064][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:14:35,746][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:26:53,279][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:39:14,878][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:51:51,180][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_FreqSelect/train_ddp_process_7.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-02 16:44:48,487][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 16:45:06,851][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 16:45:06,852][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 16:45:26,166][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 16:45:39,979][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 16:45:40,602][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 16:47:13,415][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-02 16:58:03,116][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:10:31,606][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:22:56,474][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:35:19,469][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 17:47:38,497][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:00:01,397][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:12:22,307][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:24:57,886][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:37:24,971][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 18:49:44,522][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:02:07,066][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:14:35,746][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:26:53,279][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:39:14,878][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:51:51,181][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_FreqSelect/wandb/debug-internal.log ADDED Viewed

	@@ -0,0 +1,12 @@

+{"time":"2026-03-02T16:45:22.318370129Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-02T16:45:22.758255134Z","level":"INFO","msg":"stream: created new stream","id":"8h1p5i95"}
+{"time":"2026-03-02T16:45:22.758429796Z","level":"INFO","msg":"handler: started","stream_id":"8h1p5i95"}
+{"time":"2026-03-02T16:45:22.7588083Z","level":"INFO","msg":"stream: started","id":"8h1p5i95"}
+{"time":"2026-03-02T16:45:22.758888841Z","level":"INFO","msg":"writer: started","stream_id":"8h1p5i95"}
+{"time":"2026-03-02T16:45:22.758889041Z","level":"INFO","msg":"sender: started","stream_id":"8h1p5i95"}
+{"time":"2026-03-02T18:53:53.537017028Z","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/know/DCSplat/8h1p5i95/file_stream","body":"\n<html><head>\n<meta http-equiv=\"content-type\" content=\"text/html;charset=utf-8\">\n<title>502 Server Error</title>\n</head>\n<body text=#000000 bgcolor=#ffffff>\n<h1>Error: Server Error</h1>\n<h2>The server encountered a temporary error and could not complete your request.<p>Please try again in 30 seconds.</h2>\n<h2></h2>\n</body></html>\n"}
+{"time":"2026-03-02T19:52:00.27915096Z","level":"INFO","msg":"stream: closing","id":"8h1p5i95"}
+{"time":"2026-03-02T19:52:01.130417487Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-03-02T19:52:01.649912042Z","level":"INFO","msg":"handler: closed","stream_id":"8h1p5i95"}
+{"time":"2026-03-02T19:52:01.650177675Z","level":"INFO","msg":"sender: closed","stream_id":"8h1p5i95"}
+{"time":"2026-03-02T19:52:01.650209056Z","level":"INFO","msg":"stream: closed","id":"8h1p5i95"}

ABLATION_0302_FreqSelect/wandb/debug.log ADDED Viewed

	@@ -0,0 +1,21 @@

+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_setup.py:_flush():81] Current SDK version is 0.25.0
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_setup.py:_flush():81] Configure stats pid to 766297
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_init.py:setup_run_log_directory():717] Logging user logs to /workspace/code/CVPR2026/outputs/ablation/re10k/ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/logs/debug.log
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_init.py:setup_run_log_directory():718] Logging internal logs to /workspace/code/CVPR2026/outputs/ablation/re10k/ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/logs/debug-internal.log
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_init.py:init():844] calling init triggers
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_init.py:init():849] wandb.init called with sweep_config: {}
+config: {'model': {'encoder': {'name': 'dcsplat', 'input_image_shape': [518, 518], 'head_mode': 'pcd', 'num_level': 3, 'gs_param_dim': 256, 'align_corners': False, 'use_voxelize': True}, 'decoder': {'name': 'splatting_cuda', 'background_color': [0.0, 0.0, 0.0], 'make_scale_invariant': False}, 'density_control': {'name': 'density_control_module', 'mean_dim': 32, 'gs_param_dim': 256, 'refinement_layer_num': 1, 'num_level': 3, 'grad_mode': 'absgrad', 'use_mean_features': True, 'refinement_type': 'voxelize', 'refinement_hidden_dim': 32, 'aggregation_mode': 'mean', 'num_heads': 1, 'score_mode': 'frequency', 'latent_dim': 128, 'num_latents': 64, 'num_self_attn_per_block': 2, 'voxel_size': 0.001, 'aux_refine': False, 'refine_error': False, 'use_refine_module': False, 'voxelize_activate': False, 'use_depth': False}}, 'render_loss': {'mse': {'weight': 1.0}, 'lpips': {'weight': 0.05, 'apply_after_step': 0}}, 'density_control_loss': {'error_score': {'weight': 0.0001, 'log_scale': False, 'grad_scale': 10000.0, 'mode': 'original'}}, 'direct_loss': {'l1': {'weight': 0.8}, 'ssim': {'weight': 0.2}}, 'wandb': {'project': 'DCSplat', 'entity': 'scene-representation-group', 'name': 'ABLATION_0302_FreqSelect', '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': 25, 'backbone_lr_multiplier': 0.1, 'backbone_trainable': 'T+H', 'accumulate': 1}, 'checkpointing': {'load': None, 'every_n_train_steps': 1500, 'save_top_k': 2, 'save_weights_only': False}, 'train': {'extended_visualization': False, 'print_log_every_n_steps': 10, 'camera_loss': 10.0, 'one_sample_validation': None, 'align_corners': False, 'intrinsic_scaling': False, 'verbose': False, 'beta_dist_param': [0.5, 4.0], 'use_refine_aux': False, 'train_target_set': True, 'train_gs_num': 1, 'ext_scale_detach': False, 'cam_scale_mode': 'sum', 'scene_scale_reg_loss': 0.01, 'train_aux': True, 'vggt_cam_loss': True, 'vggt_distil': False, 'context_view_train': False}, 'test': {'output_path': 'test/ablation/re10k', '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_active_mask_image': False, 'save_error_score_image': False, 'save_compare': False, 'save_gs': False, 'save_sample_wise_metrics': True, 'pred_intrinsic': False, 'error_threshold': 0.4, 'error_threshold_list': [0.2, 0.4, 0.6, 0.8, 1.0], 'threshold_mode': 'ratio', 'nvs_view_N_list': [3, 6, 16, 32, 64]}, 'seed': 111123, 'trainer': {'max_steps': 3001, 'val_check_interval': 250, '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': 1000, 'initial_min_distance_between_context_views': 25, 'initial_max_distance_between_context_views': 25, 'same_target_gap': False, 'num_target_set': 3, 'target_align': True}, '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, 'dynamic_context_views': True, 'max_context_views_per_gpu': 24}}, '_wandb': {}}
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_init.py:init():892] starting backend
+2026-03-02 16:45:22,307 INFO    MainThread:766297 [wandb_init.py:init():895] sending inform_init request
+2026-03-02 16:45:22,314 INFO    MainThread:766297 [wandb_init.py:init():903] backend started and connected
+2026-03-02 16:45:22,319 INFO    MainThread:766297 [wandb_init.py:init():973] updated telemetry
+2026-03-02 16:45:22,328 INFO    MainThread:766297 [wandb_init.py:init():997] communicating run to backend with 90.0 second timeout
+2026-03-02 16:45:23,308 INFO    MainThread:766297 [wandb_init.py:init():1042] starting run threads in backend
+2026-03-02 16:45:23,389 INFO    MainThread:766297 [wandb_run.py:_console_start():2524] atexit reg
+2026-03-02 16:45:23,389 INFO    MainThread:766297 [wandb_run.py:_redirect():2373] redirect: wrap_raw
+2026-03-02 16:45:23,389 INFO    MainThread:766297 [wandb_run.py:_redirect():2442] Wrapping output streams.
+2026-03-02 16:45:23,389 INFO    MainThread:766297 [wandb_run.py:_redirect():2465] Redirects installed.
+2026-03-02 16:45:23,392 INFO    MainThread:766297 [wandb_init.py:init():1082] run started, returning control to user process
+2026-03-02 19:52:00,279 INFO    wandb-AsyncioManager-main:766297 [service_client.py:_forward_responses():134] Reached EOF.
+2026-03-02 19:52:00,279 INFO    wandb-AsyncioManager-main:766297 [mailbox.py:close():155] Closing mailbox, abandoning 1 handles.

ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/files/config.yaml ADDED Viewed

	@@ -0,0 +1,310 @@

+_wandb:
+    value:
+        cli_version: 0.25.0
+        e:
+            um6r6nvkzkoatbnm303ud7x9azv8duwx:
+                args:
+                    - +experiment=re10k_ablation_24v
+                    - wandb.mode=online
+                    - wandb.name=ABLATION_0302_FreqSelect
+                    - model.density_control.score_mode=frequency
+                cpu_count: 128
+                cpu_count_logical: 256
+                cudaVersion: "13.0"
+                disk:
+                    /:
+                        total: "735513149440"
+                        used: "700478996480"
+                email: dna9041@korea.ac.kr
+                executable: /venv/main/bin/python
+                git:
+                    commit: db456c095bcedf29b387e0d292e95b859d67a57e
+                    remote: git@github.com:K-nowing/CVPR2026.git
+                gpu: NVIDIA H200
+                gpu_count: 8
+                gpu_nvidia:
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "150754820096"
+                      name: NVIDIA H200
+                      uuid: GPU-9a20101e-d876-facd-5f05-805081aede41
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "150754820096"
+                      name: NVIDIA H200
+                      uuid: GPU-84736a77-ee75-3324-e4e1-99cc15bfb5e9
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "150754820096"
+                      name: NVIDIA H200
+                      uuid: GPU-423d3161-cdc4-3fc0-caee-d15cfaa83ca6
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "150754820096"
+                      name: NVIDIA H200
+                      uuid: GPU-5b0058b2-cdb9-c952-04f9-87dcaa7ea742
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "150754820096"
+                      name: NVIDIA H200
+                      uuid: GPU-08b37f98-4603-d483-2f2b-fe5311aa42f2
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "150754820096"
+                      name: NVIDIA H200
+                      uuid: GPU-03273b5b-2fdd-a5fe-4460-c897334ae464
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "150754820096"
+                      name: NVIDIA H200
+                      uuid: GPU-292d466c-d00d-25a4-28b6-e6c978d3e70c
+                    - architecture: Hopper
+                      cudaCores: 16896
+                      memoryTotal: "150754820096"
+                      name: NVIDIA H200
+                      uuid: GPU-46f38561-3148-e442-7f7f-bfe447bab7fe
+                host: e9d3310a05da
+                memory:
+                    total: "1622950240256"
+                os: Linux-6.8.0-94-generic-x86_64-with-glibc2.39
+                program: -m src.main
+                python: CPython 3.12.12
+                root: /workspace/code/CVPR2026/outputs/ablation/re10k/ABLATION_0302_FreqSelect
+                startedAt: "2026-03-02T16:45:22.019994Z"
+                writerId: um6r6nvkzkoatbnm303ud7x9azv8duwx
+        m:
+            - "1": trainer/global_step
+              "6":
+                - 3
+              "7": []
+            - "2": '*'
+              "5": 1
+              "6":
+                - 1
+              "7": []
+        python_version: 3.12.12
+        t:
+            "1":
+                - 1
+                - 41
+                - 49
+                - 50
+                - 106
+            "2":
+                - 1
+                - 41
+                - 49
+                - 50
+                - 106
+            "3":
+                - 7
+                - 13
+                - 15
+                - 16
+                - 66
+            "4": 3.12.12
+            "5": 0.25.0
+            "12": 0.25.0
+            "13": linux-x86_64
+checkpointing:
+    value:
+        every_n_train_steps: 1500
+        load: null
+        save_top_k: 2
+        save_weights_only: false
+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
+            dynamic_context_views: true
+            input_image_shape:
+                - 256
+                - 256
+            make_baseline_1: true
+            max_context_views_per_gpu: 24
+            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:
+                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_set: 3
+                num_target_views: 4
+                same_target_gap: false
+                target_align: true
+                warm_up_steps: 1000
+density_control_loss:
+    value:
+        error_score:
+            grad_scale: 10000
+            log_scale: false
+            mode: original
+            weight: 0.0001
+direct_loss:
+    value:
+        l1:
+            weight: 0.8
+        ssim:
+            weight: 0.2
+mode:
+    value: train
+model:
+    value:
+        decoder:
+            background_color:
+                - 0
+                - 0
+                - 0
+            make_scale_invariant: false
+            name: splatting_cuda
+        density_control:
+            aggregation_mode: mean
+            aux_refine: false
+            grad_mode: absgrad
+            gs_param_dim: 256
+            latent_dim: 128
+            mean_dim: 32
+            name: density_control_module
+            num_heads: 1
+            num_latents: 64
+            num_level: 3
+            num_self_attn_per_block: 2
+            refine_error: false
+            refinement_hidden_dim: 32
+            refinement_layer_num: 1
+            refinement_type: voxelize
+            score_mode: frequency
+            use_depth: false
+            use_mean_features: true
+            use_refine_module: false
+            voxel_size: 0.001
+            voxelize_activate: false
+        encoder:
+            align_corners: false
+            gs_param_dim: 256
+            head_mode: pcd
+            input_image_shape:
+                - 518
+                - 518
+            name: dcsplat
+            num_level: 3
+            use_voxelize: true
+optimizer:
+    value:
+        accumulate: 1
+        backbone_lr_multiplier: 0.1
+        backbone_trainable: T+H
+        lr: 0.0002
+        warm_up_steps: 25
+render_loss:
+    value:
+        lpips:
+            apply_after_step: 0
+            weight: 0.05
+        mse:
+            weight: 1
+seed:
+    value: 111123
+test:
+    value:
+        align_pose: false
+        compute_scores: true
+        error_threshold: 0.4
+        error_threshold_list:
+            - 0.2
+            - 0.4
+            - 0.6
+            - 0.8
+            - 1
+        nvs_view_N_list:
+            - 3
+            - 6
+            - 16
+            - 32
+            - 64
+        output_path: test/ablation/re10k
+        pose_align_steps: 100
+        pred_intrinsic: false
+        rot_opt_lr: 0.005
+        save_active_mask_image: false
+        save_compare: false
+        save_error_score_image: false
+        save_gs: false
+        save_image: false
+        save_sample_wise_metrics: true
+        save_video: false
+        threshold_mode: ratio
+        trans_opt_lr: 0.005
+train:
+    value:
+        align_corners: false
+        beta_dist_param:
+            - 0.5
+            - 4
+        cam_scale_mode: sum
+        camera_loss: 10
+        context_view_train: false
+        ext_scale_detach: false
+        extended_visualization: false
+        intrinsic_scaling: false
+        one_sample_validation: null
+        print_log_every_n_steps: 10
+        scene_scale_reg_loss: 0.01
+        train_aux: true
+        train_gs_num: 1
+        train_target_set: true
+        use_refine_aux: false
+        verbose: false
+        vggt_cam_loss: true
+        vggt_distil: false
+trainer:
+    value:
+        gradient_clip_val: 0.5
+        max_steps: 3001
+        num_nodes: 1
+        val_check_interval: 250
+wandb:
+    value:
+        entity: scene-representation-group
+        mode: online
+        name: ABLATION_0302_FreqSelect
+        project: DCSplat
+        tags:
+            - re10k
+            - 256x256

ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/files/output.log ADDED Viewed

The diff for this file is too large to render. See raw diff

ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/files/requirements.txt ADDED Viewed

	@@ -0,0 +1,173 @@

+wheel==0.45.1
+pytz==2025.2
+easydict==1.13
+antlr4-python3-runtime==4.9.3
+wadler_lindig==0.1.7
+networkx==3.4.2
+urllib3==2.5.0
+tzdata==2025.2
+typing-inspection==0.4.1
+tabulate==0.9.0
+smmap==5.0.2
+setuptools==78.1.1
+safetensors==0.5.3
+multidict==6.6.4
+PyYAML==6.0.2
+PySocks==1.7.1
+pyparsing==3.2.5
+pydantic_core==2.33.2
+pycparser==2.23
+protobuf==6.32.1
+propcache==0.3.2
+proglog==0.1.12
+platformdirs==4.4.0
+pip==25.2
+mdurl==0.1.2
+pillow==10.4.0
+packaging==24.2
+opt_einsum==3.4.0
+frozenlist==1.7.0
+numpy==1.26.4
+ninja==1.13.0
+MarkupSafe==3.0.2
+kornia_rs==0.1.9
+kiwisolver==1.4.9
+imageio-ffmpeg==0.6.0
+idna==3.7
+fsspec==2024.6.1
+hf-xet==1.1.10
+gmpy2==2.2.1
+fonttools==4.60.0
+triton==3.4.0
+filelock==3.17.0
+einops==0.8.1
+decorator==4.4.2
+dacite==1.9.2
+cycler==0.12.1
+colorama==0.4.6
+click==8.3.0
+nvidia-nvtx-cu12==12.8.90
+charset-normalizer==3.3.2
+certifi==2025.8.3
+beartype==0.19.0
+attrs==25.3.0
+async-timeout==5.0.1
+annotated-types==0.7.0
+aiohappyeyeballs==2.6.1
+yarl==1.20.1
+tifffile==2025.5.10
+sentry-sdk==2.39.0
+scipy==1.15.3
+pydantic==2.11.9
+pandas==2.3.2
+opencv-python==4.11.0.86
+omegaconf==2.3.0
+markdown-it-py==4.0.0
+lightning-utilities==0.14.3
+lazy_loader==0.4
+jaxtyping==0.2.37
+imageio==2.37.0
+gitdb==4.0.12
+contourpy==1.3.2
+colorspacious==1.1.2
+cffi==1.17.1
+aiosignal==1.4.0
+scikit-video==1.1.11
+scikit-image==0.25.2
+rich==14.1.0
+moviepy==1.0.3
+matplotlib==3.10.6
+hydra-core==1.3.2
+huggingface-hub==0.35.1
+GitPython==3.1.45
+brotlicffi==1.0.9.2
+aiohttp==3.12.15
+torchmetrics==1.8.2
+opt-einsum-fx==0.1.4
+kornia==0.8.1
+pytorch-lightning==2.5.1
+lpips==0.1.4
+e3nn==0.6.0
+lightning==2.5.1
+gsplat==1.5.3
+nvidia-cusparselt-cu12==0.7.1
+nvidia-nvjitlink-cu12==12.8.93
+nvidia-nccl-cu12==2.27.3
+nvidia-curand-cu12==10.3.9.90
+nvidia-cufile-cu12==1.13.1.3
+nvidia-cuda-runtime-cu12==12.8.90
+nvidia-cuda-nvrtc-cu12==12.8.93
+nvidia-cuda-cupti-cu12==12.8.90
+nvidia-cublas-cu12==12.8.4.1
+nvidia-cusparse-cu12==12.5.8.93
+nvidia-cufft-cu12==11.3.3.83
+nvidia-cudnn-cu12==9.10.2.21
+nvidia-cusolver-cu12==11.7.3.90
+torch==2.8.0+cu128
+torchvision==0.23.0+cu128
+torchaudio==2.8.0+cu128
+torch_scatter==2.1.2+pt28cu128
+plyfile==1.1.3
+wandb==0.25.0
+cuda-bindings==13.0.3
+cuda-pathfinder==1.3.3
+Jinja2==3.1.6
+mpmath==1.3.0
+nvidia-cublas==13.1.0.3
+nvidia-cuda-cupti==13.0.85
+nvidia-cuda-nvrtc==13.0.88
+nvidia-cuda-runtime==13.0.96
+nvidia-cudnn-cu13==9.15.1.9
+nvidia-cufft==12.0.0.61
+nvidia-cufile==1.15.1.6
+nvidia-curand==10.4.0.35
+nvidia-cusolver==12.0.4.66
+nvidia-cusparse==12.6.3.3
+nvidia-cusparselt-cu13==0.8.0
+nvidia-nccl-cu13==2.28.9
+nvidia-nvjitlink==13.0.88
+nvidia-nvshmem-cu13==3.4.5
+nvidia-nvtx==13.0.85
+requests==2.32.5
+sentencepiece==0.2.1
+sympy==1.14.0
+torchcodec==0.10.0
+torchdata==0.10.0
+torchtext==0.6.0
+anyio==4.12.0
+asttokens==3.0.1
+comm==0.2.3
+debugpy==1.8.19
+executing==2.2.1
+h11==0.16.0
+httpcore==1.0.9
+httpx==0.28.1
+ipykernel==7.1.0
+ipython==9.8.0
+ipython_pygments_lexers==1.1.1
+ipywidgets==8.1.8
+jedi==0.19.2
+jupyter_client==8.7.0
+jupyter_core==5.9.1
+jupyterlab_widgets==3.0.16
+matplotlib-inline==0.2.1
+nest-asyncio==1.6.0
+parso==0.8.5
+pexpect==4.9.0
+prompt_toolkit==3.0.52
+psutil==7.2.1
+ptyprocess==0.7.0
+pure_eval==0.2.3
+Pygments==2.19.2
+python-dateutil==2.9.0.post0
+pyzmq==27.1.0
+shellingham==1.5.4
+six==1.17.0
+stack-data==0.6.3
+tornado==6.5.4
+tqdm==4.67.1
+traitlets==5.14.3
+typer-slim==0.21.0
+typing_extensions==4.15.0
+wcwidth==0.2.14
+widgetsnbextension==4.0.15

ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/files/wandb-metadata.json ADDED Viewed

	@@ -0,0 +1,93 @@

+{
+  "os":  "Linux-6.8.0-94-generic-x86_64-with-glibc2.39",
+  "python":  "CPython 3.12.12",
+  "startedAt":  "2026-03-02T16:45:22.019994Z",
+  "args":  [
+    "+experiment=re10k_ablation_24v",
+    "wandb.mode=online",
+    "wandb.name=ABLATION_0302_FreqSelect",
+    "model.density_control.score_mode=frequency"
+  ],
+  "program":  "-m src.main",
+  "git":  {
+    "remote":  "git@github.com:K-nowing/CVPR2026.git",
+    "commit":  "db456c095bcedf29b387e0d292e95b859d67a57e"
+  },
+  "email":  "dna9041@korea.ac.kr",
+  "root":  "/workspace/code/CVPR2026/outputs/ablation/re10k/ABLATION_0302_FreqSelect",
+  "host":  "e9d3310a05da",
+  "executable":  "/venv/main/bin/python",
+  "cpu_count":  128,
+  "cpu_count_logical":  256,
+  "gpu":  "NVIDIA H200",
+  "gpu_count":  8,
+  "disk":  {
+    "/":  {
+      "total":  "735513149440",
+      "used":  "700478996480"
+    }
+  },
+  "memory":  {
+    "total":  "1622950240256"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA H200",
+      "memoryTotal":  "150754820096",
+      "cudaCores":  16896,
+      "architecture":  "Hopper",
+      "uuid":  "GPU-9a20101e-d876-facd-5f05-805081aede41"
+    },
+    {
+      "name":  "NVIDIA H200",
+      "memoryTotal":  "150754820096",
+      "cudaCores":  16896,
+      "architecture":  "Hopper",
+      "uuid":  "GPU-84736a77-ee75-3324-e4e1-99cc15bfb5e9"
+    },
+    {
+      "name":  "NVIDIA H200",
+      "memoryTotal":  "150754820096",
+      "cudaCores":  16896,
+      "architecture":  "Hopper",
+      "uuid":  "GPU-423d3161-cdc4-3fc0-caee-d15cfaa83ca6"
+    },
+    {
+      "name":  "NVIDIA H200",
+      "memoryTotal":  "150754820096",
+      "cudaCores":  16896,
+      "architecture":  "Hopper",
+      "uuid":  "GPU-5b0058b2-cdb9-c952-04f9-87dcaa7ea742"
+    },
+    {
+      "name":  "NVIDIA H200",
+      "memoryTotal":  "150754820096",
+      "cudaCores":  16896,
+      "architecture":  "Hopper",
+      "uuid":  "GPU-08b37f98-4603-d483-2f2b-fe5311aa42f2"
+    },
+    {
+      "name":  "NVIDIA H200",
+      "memoryTotal":  "150754820096",
+      "cudaCores":  16896,
+      "architecture":  "Hopper",
+      "uuid":  "GPU-03273b5b-2fdd-a5fe-4460-c897334ae464"
+    },
+    {
+      "name":  "NVIDIA H200",
+      "memoryTotal":  "150754820096",
+      "cudaCores":  16896,
+      "architecture":  "Hopper",
+      "uuid":  "GPU-292d466c-d00d-25a4-28b6-e6c978d3e70c"
+    },
+    {
+      "name":  "NVIDIA H200",
+      "memoryTotal":  "150754820096",
+      "cudaCores":  16896,
+      "architecture":  "Hopper",
+      "uuid":  "GPU-46f38561-3148-e442-7f7f-bfe447bab7fe"
+    }
+  ],
+  "cudaVersion":  "13.0",
+  "writerId":  "um6r6nvkzkoatbnm303ud7x9azv8duwx"
+}

ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/files/wandb-summary.json ADDED Viewed

	@@ -0,0 +1 @@

+ {"val/lpips":0.15850743651390076,"train/error_scores":{"_type":"images/separated","width":1328,"height":2120,"format":"png","count":1,"filenames":["media/images/train/error_scores_201_0241694c9f816d9fdf4d.png"],"captions":[["0621c7675fab1418"]]},"train/scene_scale":1.0069646835327148,"lr-AdamW/pg1-momentum":0.9,"val/gaussian_num_ratio":0.400054931640625,"loss/scene_scale_reg":0.0001745200133882463,"_wandb":{"runtime":11196},"_timestamp":1.772481113291296e+09,"loss/aux_1/mse":0.010353680700063705,"_step":202,"_runtime":11196,"epoch":0,"loss/camera":0.0005779359489679337,"active_mask_imgs":{"filenames":["media/images/active_mask_imgs_198_24c7ded6b719c7a30450.png"],"captions":["a76028640ffa1ef9"],"_type":"images/separated","width":536,"height":800,"format":"png","count":1},"loss/aux_0/lpips":0.01015381421893835,"loss/aux_0/mse":0.009505179710686207,"lr-AdamW/pg1":2.003594834351718e-05,"train/psnr_probabilistic":20.417490005493164,"train/comparison":{"count":1,"filenames":["media/images/train/comparison_202_ca1bb2583132c1b8f525.png"],"captions":[["0621c7675fab1418"]],"_type":"images/separated","width":1328,"height":2154,"format":"png"},"info/global_step":3000,"comparison":{"height":1098,"format":"png","count":1,"filenames":["media/images/comparison_197_06ad81be46157c558162.png"],"captions":["a76028640ffa1ef9"],"_type":"images/separated","width":1064},"lr-AdamW/pg2":2e-05,"loss/aux_2/mse":0.010289942845702171,"loss/total":0.08268259465694427,"loss/aux_1/lpips":0.009140809997916222,"lr-AdamW/pg2-momentum":0.9,"loss/final_3dgs/lpips":0.008590941317379475,"trainer/global_step":3001,"val/psnr":22.65956687927246,"loss/aux_2/lpips":0.008691498078405857,"loss/aux_0/error_score":0.7447234392166138,"loss/aux_1/error_score":0.4630742371082306,"loss/final_3dgs/mse":0.010002846829593182,"error_scores":{"filenames":["media/images/error_scores_199_bbf557521907e54e9e40.png"],"captions":["a76028640ffa1ef9"],"_type":"images/separated","width":800,"height":536,"format":"png","count":1},"val/ssim":0.835202693939209}

ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/logs/debug-core.log ADDED Viewed

	@@ -0,0 +1,15 @@

+{"time":"2026-03-02T16:45:22.126555248Z","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpcn3kvkw7/port-766297.txt","pid":766297,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-03-02T16:45:22.127234006Z","level":"INFO","msg":"server: will exit if parent process dies","ppid":766297}
+{"time":"2026-03-02T16:45:22.127219646Z","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-766297-768857-588636342/socket","Net":"unix"}}
+{"time":"2026-03-02T16:45:22.306567301Z","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-03-02T16:45:22.318114736Z","level":"INFO","msg":"handleInformInit: received","streamId":"8h1p5i95","id":"1(@)"}
+{"time":"2026-03-02T16:45:22.758828431Z","level":"INFO","msg":"handleInformInit: stream started","streamId":"8h1p5i95","id":"1(@)"}
+{"time":"2026-03-02T16:45:28.487099971Z","level":"INFO","msg":"connection: cancelling request","id":"1(@)","requestId":"javd7cpr2iqo"}
+{"time":"2026-03-02T19:52:00.279083819Z","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2026-03-02T19:52:00.2791473Z","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2026-03-02T19:52:00.27916101Z","level":"INFO","msg":"server is shutting down"}
+{"time":"2026-03-02T19:52:00.279200001Z","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2026-03-02T19:52:00.279293062Z","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-766297-768857-588636342/socket","Net":"unix"}}
+{"time":"2026-03-02T19:52:01.650696431Z","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2026-03-02T19:52:01.650729252Z","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2026-03-02T19:52:01.650746272Z","level":"INFO","msg":"server is closed"}

ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/logs/debug-internal.log ADDED Viewed

	@@ -0,0 +1,12 @@

+{"time":"2026-03-02T16:45:22.318370129Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-02T16:45:22.758255134Z","level":"INFO","msg":"stream: created new stream","id":"8h1p5i95"}
+{"time":"2026-03-02T16:45:22.758429796Z","level":"INFO","msg":"handler: started","stream_id":"8h1p5i95"}
+{"time":"2026-03-02T16:45:22.7588083Z","level":"INFO","msg":"stream: started","id":"8h1p5i95"}
+{"time":"2026-03-02T16:45:22.758888841Z","level":"INFO","msg":"writer: started","stream_id":"8h1p5i95"}
+{"time":"2026-03-02T16:45:22.758889041Z","level":"INFO","msg":"sender: started","stream_id":"8h1p5i95"}
+{"time":"2026-03-02T18:53:53.537017028Z","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/know/DCSplat/8h1p5i95/file_stream","body":"\n<html><head>\n<meta http-equiv=\"content-type\" content=\"text/html;charset=utf-8\">\n<title>502 Server Error</title>\n</head>\n<body text=#000000 bgcolor=#ffffff>\n<h1>Error: Server Error</h1>\n<h2>The server encountered a temporary error and could not complete your request.<p>Please try again in 30 seconds.</h2>\n<h2></h2>\n</body></html>\n"}
+{"time":"2026-03-02T19:52:00.27915096Z","level":"INFO","msg":"stream: closing","id":"8h1p5i95"}
+{"time":"2026-03-02T19:52:01.130417487Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-03-02T19:52:01.649912042Z","level":"INFO","msg":"handler: closed","stream_id":"8h1p5i95"}
+{"time":"2026-03-02T19:52:01.650177675Z","level":"INFO","msg":"sender: closed","stream_id":"8h1p5i95"}
+{"time":"2026-03-02T19:52:01.650209056Z","level":"INFO","msg":"stream: closed","id":"8h1p5i95"}

ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/logs/debug.log ADDED Viewed

	@@ -0,0 +1,21 @@

+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_setup.py:_flush():81] Current SDK version is 0.25.0
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_setup.py:_flush():81] Configure stats pid to 766297
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_init.py:setup_run_log_directory():717] Logging user logs to /workspace/code/CVPR2026/outputs/ablation/re10k/ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/logs/debug.log
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_init.py:setup_run_log_directory():718] Logging internal logs to /workspace/code/CVPR2026/outputs/ablation/re10k/ABLATION_0302_FreqSelect/wandb/run-20260302_164522-8h1p5i95/logs/debug-internal.log
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_init.py:init():844] calling init triggers
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_init.py:init():849] wandb.init called with sweep_config: {}
+config: {'model': {'encoder': {'name': 'dcsplat', 'input_image_shape': [518, 518], 'head_mode': 'pcd', 'num_level': 3, 'gs_param_dim': 256, 'align_corners': False, 'use_voxelize': True}, 'decoder': {'name': 'splatting_cuda', 'background_color': [0.0, 0.0, 0.0], 'make_scale_invariant': False}, 'density_control': {'name': 'density_control_module', 'mean_dim': 32, 'gs_param_dim': 256, 'refinement_layer_num': 1, 'num_level': 3, 'grad_mode': 'absgrad', 'use_mean_features': True, 'refinement_type': 'voxelize', 'refinement_hidden_dim': 32, 'aggregation_mode': 'mean', 'num_heads': 1, 'score_mode': 'frequency', 'latent_dim': 128, 'num_latents': 64, 'num_self_attn_per_block': 2, 'voxel_size': 0.001, 'aux_refine': False, 'refine_error': False, 'use_refine_module': False, 'voxelize_activate': False, 'use_depth': False}}, 'render_loss': {'mse': {'weight': 1.0}, 'lpips': {'weight': 0.05, 'apply_after_step': 0}}, 'density_control_loss': {'error_score': {'weight': 0.0001, 'log_scale': False, 'grad_scale': 10000.0, 'mode': 'original'}}, 'direct_loss': {'l1': {'weight': 0.8}, 'ssim': {'weight': 0.2}}, 'wandb': {'project': 'DCSplat', 'entity': 'scene-representation-group', 'name': 'ABLATION_0302_FreqSelect', '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': 25, 'backbone_lr_multiplier': 0.1, 'backbone_trainable': 'T+H', 'accumulate': 1}, 'checkpointing': {'load': None, 'every_n_train_steps': 1500, 'save_top_k': 2, 'save_weights_only': False}, 'train': {'extended_visualization': False, 'print_log_every_n_steps': 10, 'camera_loss': 10.0, 'one_sample_validation': None, 'align_corners': False, 'intrinsic_scaling': False, 'verbose': False, 'beta_dist_param': [0.5, 4.0], 'use_refine_aux': False, 'train_target_set': True, 'train_gs_num': 1, 'ext_scale_detach': False, 'cam_scale_mode': 'sum', 'scene_scale_reg_loss': 0.01, 'train_aux': True, 'vggt_cam_loss': True, 'vggt_distil': False, 'context_view_train': False}, 'test': {'output_path': 'test/ablation/re10k', '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_active_mask_image': False, 'save_error_score_image': False, 'save_compare': False, 'save_gs': False, 'save_sample_wise_metrics': True, 'pred_intrinsic': False, 'error_threshold': 0.4, 'error_threshold_list': [0.2, 0.4, 0.6, 0.8, 1.0], 'threshold_mode': 'ratio', 'nvs_view_N_list': [3, 6, 16, 32, 64]}, 'seed': 111123, 'trainer': {'max_steps': 3001, 'val_check_interval': 250, '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': 1000, 'initial_min_distance_between_context_views': 25, 'initial_max_distance_between_context_views': 25, 'same_target_gap': False, 'num_target_set': 3, 'target_align': True}, '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, 'dynamic_context_views': True, 'max_context_views_per_gpu': 24}}, '_wandb': {}}
+2026-03-02 16:45:22,023 INFO    MainThread:766297 [wandb_init.py:init():892] starting backend
+2026-03-02 16:45:22,307 INFO    MainThread:766297 [wandb_init.py:init():895] sending inform_init request
+2026-03-02 16:45:22,314 INFO    MainThread:766297 [wandb_init.py:init():903] backend started and connected
+2026-03-02 16:45:22,319 INFO    MainThread:766297 [wandb_init.py:init():973] updated telemetry
+2026-03-02 16:45:22,328 INFO    MainThread:766297 [wandb_init.py:init():997] communicating run to backend with 90.0 second timeout
+2026-03-02 16:45:23,308 INFO    MainThread:766297 [wandb_init.py:init():1042] starting run threads in backend
+2026-03-02 16:45:23,389 INFO    MainThread:766297 [wandb_run.py:_console_start():2524] atexit reg
+2026-03-02 16:45:23,389 INFO    MainThread:766297 [wandb_run.py:_redirect():2373] redirect: wrap_raw
+2026-03-02 16:45:23,389 INFO    MainThread:766297 [wandb_run.py:_redirect():2442] Wrapping output streams.
+2026-03-02 16:45:23,389 INFO    MainThread:766297 [wandb_run.py:_redirect():2465] Redirects installed.
+2026-03-02 16:45:23,392 INFO    MainThread:766297 [wandb_init.py:init():1082] run started, returning control to user process
+2026-03-02 19:52:00,279 INFO    wandb-AsyncioManager-main:766297 [service_client.py:_forward_responses():134] Reached EOF.
+2026-03-02 19:52:00,279 INFO    wandb-AsyncioManager-main:766297 [mailbox.py:close():155] Closing mailbox, abandoning 1 handles.

ABLATION_0302_noAux/main.log ADDED Viewed

	@@ -0,0 +1,128 @@

+[2026-03-03 14:34:03,428][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 14:34:09,556][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 14:34:09,556][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 14:34:57,994][py.warnings][WARNING] - /venv/main/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=31` in the `DataLoader` to improve performance.
+[2026-03-03 14:34:57,995][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 14:35:00,447][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:35:00,456][py.warnings][WARNING] - /venv/main/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)`.
+[2026-03-03 14:35:00,457][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 14:35:00,457][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 14:35:02,137][py.warnings][WARNING] - /venv/main/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:4322.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+[2026-03-03 14:35:02,417][py.warnings][WARNING] - /venv/main/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.
+[2026-03-03 14:35:02,418][py.warnings][WARNING] - /venv/main/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.
+[2026-03-03 14:35:02,419][py.warnings][WARNING] - /venv/main/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.
+[2026-03-03 14:35:02,419][py.warnings][WARNING] - /venv/main/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.
+[2026-03-03 14:35:02,419][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('info/global_step', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+[2026-03-03 14:35:10,752][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 14:35:10,829][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:36:34,045][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 14:46:27,539][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:49:12,153][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:57:45,642][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:03:17,724][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:08:59,407][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:17:23,741][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:20:18,762][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:31:34,183][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:31:38,356][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:42:59,995][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:45:48,381][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:54:24,331][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:00:13,471][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:05:56,522][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:14:22,749][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:17:15,694][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:28:31,361][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:28:34,568][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:39:52,044][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:42:42,659][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:51:15,071][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:56:49,315][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:02:35,913][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:11:02,286][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:13:57,869][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:25:25,165][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:25:29,251][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_noAux/peak_vram_memory.json ADDED Viewed

	@@ -0,0 +1,6 @@

+{
+  "peak_memory_allocated_gb": 70.853,
+  "peak_memory_reserved_gb": 107.16,
+  "total_elapsed_hours": 2.86,
+  "mode": "train"
+}

ABLATION_0302_noAux/train_ddp_process_1.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-03 14:34:19,853][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 14:34:36,367][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 14:34:36,368][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 14:34:57,994][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 14:35:10,752][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 14:35:10,852][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:36:34,041][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 14:46:27,538][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:57:45,641][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:08:59,405][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:20:18,760][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:31:38,353][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:42:59,992][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:54:24,328][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:05:56,519][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:17:15,692][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:28:34,566][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:39:52,041][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:51:15,068][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:02:35,910][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:13:57,866][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:25:29,250][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_noAux/train_ddp_process_2.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-03 14:34:19,744][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 14:34:41,954][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 14:34:41,954][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 14:34:57,994][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 14:35:10,752][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 14:35:10,854][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:36:34,046][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 14:46:27,541][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:57:45,642][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:08:59,409][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:20:18,764][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:31:38,356][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:42:59,995][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:54:24,330][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:05:56,522][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:17:15,695][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:28:34,568][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:39:52,044][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:51:15,071][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:02:35,913][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:13:57,869][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:25:29,252][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_noAux/train_ddp_process_3.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-03 14:34:19,818][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 14:34:46,371][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 14:34:46,372][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 14:34:57,994][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 14:35:10,752][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 14:35:10,872][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:36:34,046][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 14:46:27,539][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:57:45,641][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:08:59,407][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:20:18,763][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:31:38,355][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:42:59,995][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:54:24,331][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:05:56,522][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:17:15,694][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:28:34,567][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:39:52,044][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:51:15,071][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:02:35,914][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:13:57,869][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:25:29,253][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_noAux/train_ddp_process_4.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-03 14:34:19,852][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 14:34:35,969][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 14:34:35,969][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 14:34:57,994][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 14:35:10,754][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 14:35:10,945][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:36:34,042][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 14:46:27,538][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:57:45,639][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:08:59,405][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:20:18,761][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:31:38,353][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:42:59,993][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:54:24,328][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:05:56,520][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:17:15,693][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:28:34,566][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:39:52,042][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:51:15,069][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:02:35,911][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:13:57,867][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:25:29,250][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_noAux/train_ddp_process_5.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-03 14:34:19,796][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 14:34:46,276][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 14:34:46,280][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 14:34:57,994][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 14:35:10,752][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 14:35:10,870][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:36:34,041][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 14:46:27,538][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:57:45,639][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:08:59,405][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:20:18,759][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:31:38,353][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:42:59,993][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:54:24,327][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:05:56,519][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:17:15,691][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:28:34,566][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:39:52,041][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:51:15,068][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:02:35,910][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:13:57,866][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:25:29,250][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_noAux/train_ddp_process_6.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-03 14:34:19,861][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 14:34:46,632][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 14:34:46,632][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 14:34:57,994][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 14:35:10,752][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 14:35:10,853][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:36:34,041][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 14:46:27,537][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:57:45,639][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:08:59,405][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:20:18,760][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:31:38,353][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:42:59,992][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:54:24,327][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:05:56,519][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:17:15,692][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:28:34,565][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:39:52,042][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:51:15,069][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:02:35,911][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:13:57,867][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:25:29,250][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_noAux/train_ddp_process_7.log ADDED Viewed

	@@ -0,0 +1,66 @@

+[2026-03-03 14:34:19,822][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 14:34:46,531][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 14:34:46,531][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 14:34:57,994][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 14:35:10,753][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 14:35:10,869][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:36:34,043][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 14:46:27,540][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:57:45,639][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:08:59,405][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:20:18,759][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:31:38,353][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:42:59,992][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 15:54:24,328][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:05:56,519][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:17:15,692][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:28:34,565][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:39:52,042][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 16:51:15,068][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:02:35,911][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:13:57,866][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 17:25:29,250][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_randomSelect/.hydra/config.yaml ADDED Viewed

	@@ -0,0 +1,188 @@

+model:
+  encoder:
+    name: dcsplat
+    input_image_shape:
+    - 518
+    - 518
+    head_mode: pcd
+    num_level: 3
+    gs_param_dim: 256
+    align_corners: false
+    use_voxelize: true
+  decoder:
+    name: splatting_cuda
+    background_color:
+    - 0.0
+    - 0.0
+    - 0.0
+    make_scale_invariant: false
+  density_control:
+    name: density_control_module
+    mean_dim: 32
+    gs_param_dim: 256
+    refinement_layer_num: 1
+    num_level: 3
+    grad_mode: absgrad
+    use_mean_features: true
+    refinement_type: voxelize
+    refinement_hidden_dim: 32
+    aggregation_mode: mean
+    num_heads: 1
+    score_mode: random
+    latent_dim: 128
+    num_latents: 64
+    num_self_attn_per_block: 2
+    voxel_size: 0.001
+    aux_refine: false
+    refine_error: false
+    use_refine_module: false
+    voxelize_activate: false
+    use_depth: false
+render_loss:
+  mse:
+    weight: 1.0
+  lpips:
+    weight: 0.05
+    apply_after_step: 0
+density_control_loss:
+  error_score:
+    weight: 0.0001
+    log_scale: false
+    grad_scale: 10000.0
+    mode: original
+direct_loss:
+  l1:
+    weight: 0.8
+  ssim:
+    weight: 0.2
+wandb:
+  project: DCSplat
+  entity: scene-representation-group
+  name: ABLATION_0302_randomSelect
+  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: 25
+  backbone_lr_multiplier: 0.1
+  backbone_trainable: T+H
+  accumulate: 1
+checkpointing:
+  load: null
+  every_n_train_steps: 1500
+  save_top_k: 2
+  save_weights_only: false
+train:
+  extended_visualization: false
+  print_log_every_n_steps: 10
+  camera_loss: 10.0
+  one_sample_validation: null
+  align_corners: false
+  intrinsic_scaling: false
+  verbose: false
+  beta_dist_param:
+  - 0.5
+  - 4.0
+  use_refine_aux: false
+  train_target_set: true
+  train_gs_num: 1
+  ext_scale_detach: false
+  cam_scale_mode: sum
+  scene_scale_reg_loss: 0.01
+  train_aux: true
+  vggt_cam_loss: true
+  vggt_distil: false
+  context_view_train: false
+test:
+  output_path: test/ablation/re10k
+  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_active_mask_image: false
+  save_error_score_image: false
+  save_compare: false
+  save_gs: false
+  save_sample_wise_metrics: true
+  pred_intrinsic: false
+  error_threshold: 0.4
+  error_threshold_list:
+  - 0.2
+  - 0.4
+  - 0.6
+  - 0.8
+  - 1.0
+  threshold_mode: ratio
+  nvs_view_N_list:
+  - 3
+  - 6
+  - 16
+  - 32
+  - 64
+seed: 111123
+trainer:
+  max_steps: 3001
+  val_check_interval: 250
+  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: 1000
+      initial_min_distance_between_context_views: 25
+      initial_max_distance_between_context_views: 25
+      same_target_gap: false
+      num_target_set: 3
+      target_align: true
+    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
+    dynamic_context_views: true
+    max_context_views_per_gpu: 24

ABLATION_0302_randomSelect/.hydra/hydra.yaml ADDED Viewed

	@@ -0,0 +1,165 @@

+hydra:
+  run:
+    dir: outputs/ablation/re10k/${wandb.name}
+  sweep:
+    dir: multirun/${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_ablation_24v
+    - wandb.mode=online
+    - wandb.name=ABLATION_0302_randomSelect
+    - model.density_control.score_mode=random
+  job:
+    name: main
+    chdir: null
+    override_dirname: +experiment=re10k_ablation_24v,model.density_control.score_mode=random,wandb.mode=online,wandb.name=ABLATION_0302_randomSelect
+    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: /workspace/code/CVPR2026
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /workspace/code/CVPR2026/config
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /workspace/code/CVPR2026/outputs/ablation/re10k/ABLATION_0302_randomSelect
+    choices:
+      experiment: re10k_ablation_24v
+      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/density_control: density_control_module
+      model/decoder: splatting_cuda
+      model/encoder: dcsplat
+      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

ABLATION_0302_randomSelect/.hydra/overrides.yaml ADDED Viewed

	@@ -0,0 +1,4 @@

+- +experiment=re10k_ablation_24v
+- wandb.mode=online
+- wandb.name=ABLATION_0302_randomSelect
+- model.density_control.score_mode=random

ABLATION_0302_randomSelect/main.log ADDED Viewed

	@@ -0,0 +1,175 @@

+[2026-03-02 19:52:11,110][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 19:52:17,322][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 19:52:17,322][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 19:53:07,267][py.warnings][WARNING] - /venv/main/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=31` in the `DataLoader` to improve performance.
+[2026-03-02 19:53:07,268][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 19:53:09,707][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:53:09,718][py.warnings][WARNING] - /venv/main/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)`.
+[2026-03-02 19:53:09,719][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 19:53:09,719][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 19:53:11,397][py.warnings][WARNING] - /venv/main/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:4322.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+[2026-03-02 19:53:11,698][py.warnings][WARNING] - /venv/main/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.
+[2026-03-02 19:53:11,700][py.warnings][WARNING] - /venv/main/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.
+[2026-03-02 19:53:11,700][py.warnings][WARNING] - /venv/main/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.
+[2026-03-02 19:53:11,700][py.warnings][WARNING] - /venv/main/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.
+[2026-03-02 19:53:11,701][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('info/global_step', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+[2026-03-02 19:53:20,961][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 19:53:21,050][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:54:54,511][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:26:37,197][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 11:26:43,212][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 11:26:43,212][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 11:27:32,354][py.warnings][WARNING] - /venv/main/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=31` in the `DataLoader` to improve performance.
+[2026-03-03 11:27:32,356][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 11:27:34,773][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:27:34,782][py.warnings][WARNING] - /venv/main/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)`.
+[2026-03-03 11:27:34,783][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 11:27:34,783][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 11:27:36,391][py.warnings][WARNING] - /venv/main/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:4322.)
+  return _VF.meshgrid(tensors, **kwargs)  # type: ignore[attr-defined]
+[2026-03-03 11:27:36,680][py.warnings][WARNING] - /venv/main/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.
+[2026-03-03 11:27:36,682][py.warnings][WARNING] - /venv/main/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.
+[2026-03-03 11:27:36,682][py.warnings][WARNING] - /venv/main/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.
+[2026-03-03 11:27:36,682][py.warnings][WARNING] - /venv/main/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.
+[2026-03-03 11:27:36,683][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/lightning/pytorch/trainer/connectors/logger_connector/result.py:434: It is recommended to use `self.log('info/global_step', ..., sync_dist=True)` when logging on epoch level in distributed setting to accumulate the metric across devices.
+[2026-03-03 11:27:45,633][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 11:27:45,725][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:29:18,420][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:40:04,524][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:43:13,458][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:52:31,300][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:58:40,265][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:04:54,125][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:14:05,645][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:17:15,917][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:29:31,187][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:29:34,724][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:41:56,951][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:45:01,545][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:54:18,337][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:00:41,367][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:06:51,811][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:16:11,622][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:19:20,089][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:31:34,567][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:31:38,845][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:44:01,296][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:47:10,009][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:56:28,811][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:02:34,598][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:08:47,969][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:17:59,009][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:21:08,716][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:33:38,976][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:33:42,549][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_randomSelect/peak_vram_memory.json ADDED Viewed

	@@ -0,0 +1,6 @@

+{
+  "peak_memory_allocated_gb": 95.866,
+  "peak_memory_reserved_gb": 106.334,
+  "total_elapsed_hours": 3.12,
+  "mode": "train"
+}

ABLATION_0302_randomSelect/train_ddp_process_1.log ADDED Viewed

	@@ -0,0 +1,87 @@

+[2026-03-02 19:52:27,986][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 19:52:56,075][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 19:52:56,075][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 19:53:07,268][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 19:53:20,961][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 19:53:21,069][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:54:54,257][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:26:53,897][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 11:27:20,658][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 11:27:20,659][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 11:27:32,355][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 11:27:45,632][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 11:27:45,740][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:29:18,422][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:40:04,524][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:52:31,301][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:04:54,123][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:17:15,917][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:29:34,724][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:41:56,949][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:54:18,334][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:06:51,809][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:19:20,089][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:31:38,844][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:44:01,295][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:56:28,810][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:08:47,968][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:21:08,716][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:33:42,549][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_randomSelect/train_ddp_process_2.log ADDED Viewed

	@@ -0,0 +1,87 @@

+[2026-03-02 19:52:27,912][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 19:52:50,765][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 19:52:50,765][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 19:53:07,268][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 19:53:20,961][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 19:53:21,067][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:54:54,258][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:26:53,655][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 11:27:20,702][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 11:27:20,702][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 11:27:32,355][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 11:27:45,640][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 11:27:45,741][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:29:18,421][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:40:04,524][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:52:31,300][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:04:54,123][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:17:15,917][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:29:34,725][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:41:56,949][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:54:18,334][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:06:51,809][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:19:20,089][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:31:38,844][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:44:01,295][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:56:28,810][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:08:47,968][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:21:08,715][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:33:42,549][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_randomSelect/train_ddp_process_3.log ADDED Viewed

	@@ -0,0 +1,87 @@

+[2026-03-02 19:52:27,977][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 19:52:56,051][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 19:52:56,051][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 19:53:07,268][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 19:53:20,968][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 19:53:21,075][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:54:54,283][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:26:53,711][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 11:27:10,011][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 11:27:10,011][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 11:27:32,355][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 11:27:45,640][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 11:27:45,756][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:29:18,445][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:40:04,525][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:52:31,301][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:04:54,125][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:17:15,917][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:29:34,724][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:41:56,950][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:54:18,334][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:06:51,809][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:19:20,089][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:31:38,843][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:44:01,295][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:56:28,811][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:08:47,969][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:21:08,716][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:33:42,549][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_randomSelect/train_ddp_process_4.log ADDED Viewed

	@@ -0,0 +1,87 @@

+[2026-03-02 19:52:27,941][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 19:52:56,205][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 19:52:56,206][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 19:53:07,268][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 19:53:20,960][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 19:53:21,067][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:54:54,274][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:26:53,673][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 11:27:20,585][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 11:27:20,586][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 11:27:32,355][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 11:27:45,632][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 11:27:45,801][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:29:18,440][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:40:04,525][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:52:31,300][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:04:54,123][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:17:15,918][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:29:34,724][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:41:56,949][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:54:18,334][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:06:51,810][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:19:20,089][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:31:38,843][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:44:01,295][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:56:28,810][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:08:47,968][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:21:08,716][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:33:42,549][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_randomSelect/train_ddp_process_5.log ADDED Viewed

	@@ -0,0 +1,87 @@

+[2026-03-02 19:52:27,948][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 19:52:56,121][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 19:52:56,122][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 19:53:07,268][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 19:53:20,224][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 19:53:21,073][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:54:54,275][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:26:53,644][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 11:27:17,296][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 11:27:17,297][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 11:27:32,355][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 11:27:44,900][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 11:27:45,748][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:29:18,438][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:40:04,524][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:52:31,300][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:04:54,123][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:17:15,917][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:29:34,724][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:41:56,949][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:54:18,333][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:06:51,809][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:19:20,089][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:31:38,843][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:44:01,295][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:56:28,811][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:08:47,968][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:21:08,716][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:33:42,549][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_randomSelect/train_ddp_process_6.log ADDED Viewed

	@@ -0,0 +1,87 @@

+[2026-03-02 19:52:28,051][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 19:52:46,104][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 19:52:46,104][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 19:53:07,268][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 19:53:20,449][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 19:53:21,068][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:54:54,257][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:26:53,682][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 11:27:09,379][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 11:27:09,379][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 11:27:32,355][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 11:27:45,118][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 11:27:45,740][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:29:18,421][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:40:04,524][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:52:31,300][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:04:54,123][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:17:15,917][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:29:34,724][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:41:56,949][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:54:18,338][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:06:51,809][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:19:20,089][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:31:38,843][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:44:01,295][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:56:28,810][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:08:47,968][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:21:08,715][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:33:42,550][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_randomSelect/train_ddp_process_7.log ADDED Viewed

	@@ -0,0 +1,87 @@

+[2026-03-02 19:52:27,991][dinov2][INFO] - using MLP layer as FFN
+[2026-03-02 19:52:45,997][py.warnings][WARNING] - /venv/main/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(
+[2026-03-02 19:52:45,997][py.warnings][WARNING] - /venv/main/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)
+[2026-03-02 19:53:07,268][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-02 19:53:20,465][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-02 19:53:21,066][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-02 19:54:54,258][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:26:53,620][dinov2][INFO] - using MLP layer as FFN
+[2026-03-03 11:27:20,381][py.warnings][WARNING] - /venv/main/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(
+[2026-03-03 11:27:20,381][py.warnings][WARNING] - /venv/main/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)
+[2026-03-03 11:27:32,355][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/distributed/distributed_c10d.py:4807: 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
+[2026-03-03 11:27:45,130][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/autograd/graph.py:829: 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() = [57, 256, 1, 1], strides() = [256, 1, 256, 256]
+bucket_view.sizes() = [57, 256, 1, 1], strides() = [256, 1, 1, 1] (Triggered internally at /pytorch/torch/csrc/distributed/c10d/reducer.cpp:334.)
+  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+[2026-03-03 11:27:45,808][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:29:18,420][py.warnings][WARNING] - /venv/main/lib/python3.12/site-packages/torch/optim/lr_scheduler.py:209: 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)
+[2026-03-03 11:40:04,524][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 11:52:31,300][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:04:54,124][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:17:15,918][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:29:34,725][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:41:56,949][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 12:54:18,334][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:06:51,809][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:19:20,089][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:31:38,844][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:44:01,295][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 13:56:28,810][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:08:47,968][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:21:08,716][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay
+[2026-03-03 14:33:42,549][py.warnings][WARNING] - /workspace/code/CVPR2026/src/visualization/layout.py:105: UserWarning: Using a non-tuple sequence for multidimensional indexing is deprecated and will be changed in pytorch 2.9; use x[tuple(seq)] instead of x[seq]. In pytorch 2.9 this will be interpreted as tensor index, x[torch.tensor(seq)], which will result either in an error or a different result (Triggered internally at /pytorch/torch/csrc/autograd/python_variable_indexing.cpp:316.)
+  result[selector] = overlay

ABLATION_0302_randomSelect/wandb/debug-internal.log ADDED Viewed

	@@ -0,0 +1,11 @@

+{"time":"2026-03-03T11:27:28.036351234Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-03T11:27:28.48779323Z","level":"INFO","msg":"stream: created new stream","id":"rtvf5rqa"}
+{"time":"2026-03-03T11:27:28.487993623Z","level":"INFO","msg":"handler: started","stream_id":"rtvf5rqa"}
+{"time":"2026-03-03T11:27:28.488310336Z","level":"INFO","msg":"stream: started","id":"rtvf5rqa"}
+{"time":"2026-03-03T11:27:28.488391967Z","level":"INFO","msg":"writer: started","stream_id":"rtvf5rqa"}
+{"time":"2026-03-03T11:27:28.488406488Z","level":"INFO","msg":"sender: started","stream_id":"rtvf5rqa"}
+{"time":"2026-03-03T14:33:51.622674985Z","level":"INFO","msg":"stream: closing","id":"rtvf5rqa"}
+{"time":"2026-03-03T14:33:52.531462775Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-03-03T14:33:52.768474669Z","level":"INFO","msg":"handler: closed","stream_id":"rtvf5rqa"}
+{"time":"2026-03-03T14:33:52.76869091Z","level":"INFO","msg":"sender: closed","stream_id":"rtvf5rqa"}
+{"time":"2026-03-03T14:33:52.76871017Z","level":"INFO","msg":"stream: closed","id":"rtvf5rqa"}

ABLATION_0302_randomSelect/wandb/debug.log ADDED Viewed

	@@ -0,0 +1,21 @@

+2026-03-03 11:27:27,779 INFO    MainThread:844222 [wandb_setup.py:_flush():81] Current SDK version is 0.25.0
+2026-03-03 11:27:27,779 INFO    MainThread:844222 [wandb_setup.py:_flush():81] Configure stats pid to 844222
+2026-03-03 11:27:27,779 INFO    MainThread:844222 [wandb_setup.py:_flush():81] Loading settings from environment variables
+2026-03-03 11:27:27,779 INFO    MainThread:844222 [wandb_init.py:setup_run_log_directory():717] Logging user logs to /workspace/code/CVPR2026/outputs/ablation/re10k/ABLATION_0302_randomSelect/wandb/run-20260303_112727-rtvf5rqa/logs/debug.log
+2026-03-03 11:27:27,779 INFO    MainThread:844222 [wandb_init.py:setup_run_log_directory():718] Logging internal logs to /workspace/code/CVPR2026/outputs/ablation/re10k/ABLATION_0302_randomSelect/wandb/run-20260303_112727-rtvf5rqa/logs/debug-internal.log
+2026-03-03 11:27:27,779 INFO    MainThread:844222 [wandb_init.py:init():844] calling init triggers
+2026-03-03 11:27:27,779 INFO    MainThread:844222 [wandb_init.py:init():849] wandb.init called with sweep_config: {}
+config: {'model': {'encoder': {'name': 'dcsplat', 'input_image_shape': [518, 518], 'head_mode': 'pcd', 'num_level': 3, 'gs_param_dim': 256, 'align_corners': False, 'use_voxelize': True}, 'decoder': {'name': 'splatting_cuda', 'background_color': [0.0, 0.0, 0.0], 'make_scale_invariant': False}, 'density_control': {'name': 'density_control_module', 'mean_dim': 32, 'gs_param_dim': 256, 'refinement_layer_num': 1, 'num_level': 3, 'grad_mode': 'absgrad', 'use_mean_features': True, 'refinement_type': 'voxelize', 'refinement_hidden_dim': 32, 'aggregation_mode': 'mean', 'num_heads': 1, 'score_mode': 'random', 'latent_dim': 128, 'num_latents': 64, 'num_self_attn_per_block': 2, 'voxel_size': 0.001, 'aux_refine': False, 'refine_error': False, 'use_refine_module': False, 'voxelize_activate': False, 'use_depth': False}}, 'render_loss': {'mse': {'weight': 1.0}, 'lpips': {'weight': 0.05, 'apply_after_step': 0}}, 'density_control_loss': {'error_score': {'weight': 0.0001, 'log_scale': False, 'grad_scale': 10000.0, 'mode': 'original'}}, 'direct_loss': {'l1': {'weight': 0.8}, 'ssim': {'weight': 0.2}}, 'wandb': {'project': 'DCSplat', 'entity': 'scene-representation-group', 'name': 'ABLATION_0302_randomSelect', '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': 25, 'backbone_lr_multiplier': 0.1, 'backbone_trainable': 'T+H', 'accumulate': 1}, 'checkpointing': {'load': None, 'every_n_train_steps': 1500, 'save_top_k': 2, 'save_weights_only': False}, 'train': {'extended_visualization': False, 'print_log_every_n_steps': 10, 'camera_loss': 10.0, 'one_sample_validation': None, 'align_corners': False, 'intrinsic_scaling': False, 'verbose': False, 'beta_dist_param': [0.5, 4.0], 'use_refine_aux': False, 'train_target_set': True, 'train_gs_num': 1, 'ext_scale_detach': False, 'cam_scale_mode': 'sum', 'scene_scale_reg_loss': 0.01, 'train_aux': True, 'vggt_cam_loss': True, 'vggt_distil': False, 'context_view_train': False}, 'test': {'output_path': 'test/ablation/re10k', '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_active_mask_image': False, 'save_error_score_image': False, 'save_compare': False, 'save_gs': False, 'save_sample_wise_metrics': True, 'pred_intrinsic': False, 'error_threshold': 0.4, 'error_threshold_list': [0.2, 0.4, 0.6, 0.8, 1.0], 'threshold_mode': 'ratio', 'nvs_view_N_list': [3, 6, 16, 32, 64]}, 'seed': 111123, 'trainer': {'max_steps': 3001, 'val_check_interval': 250, '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': 1000, 'initial_min_distance_between_context_views': 25, 'initial_max_distance_between_context_views': 25, 'same_target_gap': False, 'num_target_set': 3, 'target_align': True}, '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, 'dynamic_context_views': True, 'max_context_views_per_gpu': 24}}, '_wandb': {}}
+2026-03-03 11:27:27,779 INFO    MainThread:844222 [wandb_init.py:init():892] starting backend
+2026-03-03 11:27:28,025 INFO    MainThread:844222 [wandb_init.py:init():895] sending inform_init request
+2026-03-03 11:27:28,032 INFO    MainThread:844222 [wandb_init.py:init():903] backend started and connected
+2026-03-03 11:27:28,041 INFO    MainThread:844222 [wandb_init.py:init():973] updated telemetry
+2026-03-03 11:27:28,049 INFO    MainThread:844222 [wandb_init.py:init():997] communicating run to backend with 90.0 second timeout
+2026-03-03 11:27:29,753 INFO    MainThread:844222 [wandb_init.py:init():1042] starting run threads in backend
+2026-03-03 11:27:29,881 INFO    MainThread:844222 [wandb_run.py:_console_start():2524] atexit reg
+2026-03-03 11:27:29,881 INFO    MainThread:844222 [wandb_run.py:_redirect():2373] redirect: wrap_raw
+2026-03-03 11:27:29,881 INFO    MainThread:844222 [wandb_run.py:_redirect():2442] Wrapping output streams.
+2026-03-03 11:27:29,881 INFO    MainThread:844222 [wandb_run.py:_redirect():2465] Redirects installed.
+2026-03-03 11:27:29,885 INFO    MainThread:844222 [wandb_init.py:init():1082] run started, returning control to user process
+2026-03-03 14:33:51,622 INFO    wandb-AsyncioManager-main:844222 [service_client.py:_forward_responses():134] Reached EOF.
+2026-03-03 14:33:51,622 INFO    wandb-AsyncioManager-main:844222 [mailbox.py:close():155] Closing mailbox, abandoning 1 handles.

ABLATION_0302_randomSelect/wandb/run-20260302_195303-hteryqzp/logs/debug-internal.log ADDED Viewed

	@@ -0,0 +1,48 @@

+{"time":"2026-03-02T19:53:03.325026325Z","level":"INFO","msg":"stream: starting","core version":"0.25.0"}
+{"time":"2026-03-02T19:53:03.752393391Z","level":"INFO","msg":"stream: created new stream","id":"hteryqzp"}
+{"time":"2026-03-02T19:53:03.752850706Z","level":"INFO","msg":"handler: started","stream_id":"hteryqzp"}
+{"time":"2026-03-02T19:53:03.753079209Z","level":"INFO","msg":"stream: started","id":"hteryqzp"}
+{"time":"2026-03-02T19:53:03.753091709Z","level":"INFO","msg":"writer: started","stream_id":"hteryqzp"}
+{"time":"2026-03-02T19:53:03.753110719Z","level":"INFO","msg":"sender: started","stream_id":"hteryqzp"}
+{"time":"2026-03-02T19:56:47.771287149Z","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":340}
+{"time":"2026-03-02T19:56:48.223083438Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.223417191Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.223756385Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.223766545Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.223795136Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.223836146Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.224300581Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.224349992Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.225243702Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.225272932Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.225285372Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.225294543Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.225299233Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.225303103Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.225711037Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.225725747Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.225791888Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.225796328Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.225802888Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.225807348Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.225850439Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.225863409Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.226163752Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.226180132Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.226573577Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.226583517Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.226587497Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.226594047Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.226754339Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.226758569Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.226764769Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.226774389Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.226780519Z","level":"ERROR","msg":"sender: sendSummary called after exit"}
+{"time":"2026-03-02T19:56:48.226784269Z","level":"WARN","msg":"sender: received Exit record more than once, ignoring"}
+{"time":"2026-03-02T19:56:48.226789179Z","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":34}
+{"time":"2026-03-02T19:56:48.648767143Z","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-03-02T19:56:48.889720962Z","level":"INFO","msg":"handler: operation stats","stats":{}}
+{"time":"2026-03-02T19:56:48.893634076Z","level":"INFO","msg":"stream: closing","id":"hteryqzp"}
+{"time":"2026-03-02T19:56:48.893654506Z","level":"INFO","msg":"handler: closed","stream_id":"hteryqzp"}
+{"time":"2026-03-02T19:56:48.893728077Z","level":"INFO","msg":"sender: closed","stream_id":"hteryqzp"}
+{"time":"2026-03-02T19:56:48.893740387Z","level":"INFO","msg":"stream: closed","id":"hteryqzp"}

ABLATION_0302_randomSelect/wandb/run-20260302_195303-hteryqzp/logs/debug.log ADDED Viewed

The diff for this file is too large to render. See raw diff

re10k/.hydra/config.yaml ADDED Viewed

	@@ -0,0 +1,186 @@

+model:
+  encoder:
+    name: dcsplat
+    input_image_shape:
+    - 518
+    - 518
+    head_mode: pcd
+    num_level: 3
+    gs_param_dim: 256
+    align_corners: false
+    use_voxelize: true
+  decoder:
+    name: splatting_cuda
+    background_color:
+    - 0.0
+    - 0.0
+    - 0.0
+    make_scale_invariant: false
+  density_control:
+    name: density_control_module
+    mean_dim: 32
+    gs_param_dim: 256
+    refinement_layer_num: 1
+    num_level: 3
+    grad_mode: absgrad
+    use_mean_features: true
+    refinement_type: voxelize
+    refinement_hidden_dim: 32
+    aggregation_mode: mean
+    num_heads: 1
+    score_mode: absgrad
+    latent_dim: 128
+    num_latents: 64
+    num_self_attn_per_block: 2
+    voxel_size: 0.001
+    aux_refine: false
+    refine_error: false
+    use_refine_module: false
+    voxelize_activate: true
+    use_depth: false
+render_loss:
+  mse:
+    weight: 1.0
+  lpips:
+    weight: 0.05
+    apply_after_step: 0
+density_control_loss:
+  error_score:
+    weight: 0.0001
+    log_scale: false
+    grad_scale: 10000.0
+    mode: original
+direct_loss:
+  l1:
+    weight: 0.8
+  ssim:
+    weight: 0.2
+wandb:
+  project: DCSplat
+  entity: scene-representation-group
+  name: re10k
+  mode: disabled
+  tags:
+  - re10k
+  - 256x256
+mode: test
+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: 25
+  backbone_lr_multiplier: 0.1
+  backbone_trainable: T+H
+  accumulate: 1
+checkpointing:
+  load: outputs/ablation/re10k/ABLATION_0301_2view_noRefineModule_nearFix/checkpoints/epoch_0-step_3000.ckpt
+  every_n_train_steps: 1500
+  save_top_k: 1
+  save_weights_only: false
+train:
+  extended_visualization: false
+  print_log_every_n_steps: 10
+  camera_loss: 10.0
+  one_sample_validation: null
+  align_corners: false
+  intrinsic_scaling: false
+  verbose: false
+  beta_dist_param:
+  - 0.5
+  - 4.0
+  use_refine_aux: false
+  train_target_set: true
+  train_gs_num: 1
+  ext_scale_detach: false
+  cam_scale_mode: sum
+  scene_scale_reg_loss: 0.01
+  train_aux: true
+  vggt_cam_loss: true
+  vggt_distil: false
+  context_view_train: false
+test:
+  output_path: eval_results/RE10K/ablation/16view/ABLATION_0301_2view_noRefineModule_nearFix
+  align_pose: true
+  pose_align_steps: 100
+  rot_opt_lr: 0.005
+  trans_opt_lr: 0.005
+  compute_scores: true
+  save_image: false
+  save_video: false
+  save_active_mask_image: false
+  save_error_score_image: false
+  save_compare: true
+  save_gs: false
+  save_sample_wise_metrics: true
+  pred_intrinsic: false
+  error_threshold: 0.4
+  error_threshold_list:
+  - 0.2
+  - 0.4
+  - 0.6
+  - 0.8
+  - 1.0
+  threshold_mode: ratio
+  nvs_view_N_list:
+  - 3
+  - 6
+  - 16
+  - 32
+  - 64
+seed: 111123
+trainer:
+  max_steps: 3001
+  val_check_interval: 250
+  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: evaluation
+      index_path: assets/evaluation_index_re10k_ctx_16.json
+      num_context_views: 2
+      num_target_views: 4
+      warm_up_steps: 1000
+      min_distance_between_context_views: 45
+      max_distance_between_context_views: 90
+      min_distance_to_context_views: 0
+      initial_min_distance_between_context_views: 25
+      initial_max_distance_between_context_views: 25
+    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
+    dynamic_context_views: false
+    max_context_views_per_gpu: 16

re10k/.hydra/hydra.yaml ADDED Viewed

	@@ -0,0 +1,173 @@

+hydra:
+  run:
+    dir: outputs/ablation/re10k/${wandb.name}
+  sweep:
+    dir: multirun/${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_ablation_2v
+    - mode=test
+    - wandb.mode=disabled
+    - dataset/view_sampler@dataset.re10k.view_sampler=evaluation
+    - dataset.re10k.view_sampler.index_path=assets/evaluation_index_re10k_ctx_16.json
+    - test.output_path=eval_results/RE10K/ablation/16view/ABLATION_0301_2view_noRefineModule_nearFix
+    - model.density_control.use_refine_module=false
+    - checkpointing.load=outputs/ablation/re10k/ABLATION_0301_2view_noRefineModule_nearFix/checkpoints/epoch_0-step_3000.ckpt
+    - test.save_image=false
+    - test.align_pose=true
+    - test.save_compare=true
+    - test.save_gs=false
+  job:
+    name: main
+    chdir: null
+    override_dirname: +experiment=re10k_ablation_2v,checkpointing.load=outputs/ablation/re10k/ABLATION_0301_2view_noRefineModule_nearFix/checkpoints/epoch_0-step_3000.ckpt,dataset.re10k.view_sampler.index_path=assets/evaluation_index_re10k_ctx_16.json,dataset/view_sampler@dataset.re10k.view_sampler=evaluation,mode=test,model.density_control.use_refine_module=false,test.align_pose=true,test.output_path=eval_results/RE10K/ablation/16view/ABLATION_0301_2view_noRefineModule_nearFix,test.save_compare=true,test.save_gs=false,test.save_image=false,wandb.mode=disabled
+    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: /workspace/code/CVPR2026
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /workspace/code/CVPR2026/config
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /workspace/code/CVPR2026/outputs/ablation/re10k/re10k
+    choices:
+      experiment: re10k_ablation_2v
+      dataset@dataset.re10k: re10k
+      dataset/view_sampler_dataset_specific_config@dataset.re10k.view_sampler: bounded_re10k
+      dataset/view_sampler@dataset.re10k.view_sampler: evaluation
+      model/density_control: density_control_module
+      model/decoder: splatting_cuda
+      model/encoder: dcsplat
+      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

re10k/.hydra/overrides.yaml ADDED Viewed

	@@ -0,0 +1,12 @@

+- +experiment=re10k_ablation_2v
+- mode=test
+- wandb.mode=disabled
+- dataset/view_sampler@dataset.re10k.view_sampler=evaluation
+- dataset.re10k.view_sampler.index_path=assets/evaluation_index_re10k_ctx_16.json
+- test.output_path=eval_results/RE10K/ablation/16view/ABLATION_0301_2view_noRefineModule_nearFix
+- model.density_control.use_refine_module=false
+- checkpointing.load=outputs/ablation/re10k/ABLATION_0301_2view_noRefineModule_nearFix/checkpoints/epoch_0-step_3000.ckpt
+- test.save_image=false
+- test.align_pose=true
+- test.save_compare=true
+- test.save_gs=false

re10k/main.log ADDED Viewed

The diff for this file is too large to render. See raw diff