config.json

{
  "architectures": [
    "EgoSieveModel"
  ],
  "auto_map": {
    "AutoConfig": "configuration_egosieve.EgoSieveConfig",
    "AutoModelForVideoClassification": "modeling_egosieve.EgoSieveModel"
  },
  "backbone_model_id": "facebook/dinov2-small",
  "backbone_revision": "ed25f3a31f01632728cabb09d1542f84ab7b0056",
  "boundary_labels": [
    "start",
    "end"
  ],
  "boundary_loss_weight": 1.0,
  "boundary_pos_weight": [
    11.0,
    11.0
  ],
  "calibration_source": "grouped-validation-split",
  "compiler_thresholds": {
    "enter_threshold": 0.39999999999999997,
    "exit_threshold": 0.24999999999999997,
    "include_discard": false,
    "merge_gap_s": 0.5,
    "min_duration_s": 1.0,
    "short_segment_route": "discard",
    "uncertainty_route": "review",
    "uncertainty_threshold": 0.65
  },
  "dropout": 0.15,
  "dtype": "float32",
  "hidden_size": 256,
  "id2label": {
    "0": "KEEP",
    "1": "REVIEW",
    "2": "REJECT"
  },
  "ignore_index": -100,
  "initializer_range": 0.02,
  "issue_labels": [
    "acting_hand_not_visible",
    "low_hand_activity",
    "camera_instability",
    "blur",
    "exposure",
    "scene_cut",
    "duplicate_frames"
  ],
  "issue_loss_weight": 1.0,
  "issue_pos_weight": [
    4.4,
    2.9338842975206614,
    1.0,
    1.0,
    1.0,
    1.0,
    1.0
  ],
  "issue_thresholds": {
    "acting_hand_not_visible": 0.3545205891132355,
    "blur": 0.6283682584762573,
    "camera_instability": 0.47812360525131226,
    "duplicate_frames": 0.34080421924591064,
    "exposure": 0.6307288408279419,
    "low_hand_activity": 0.3466801941394806,
    "scene_cut": 0.4578688144683838
  },
  "label2id": {
    "KEEP": 0,
    "REJECT": 2,
    "REVIEW": 1
  },
  "max_frames": 512,
  "model_type": "egosieve",
  "num_boundary_labels": 2,
  "num_frames": 12,
  "num_issue_labels": 7,
  "num_readiness_labels": 3,
  "projection_dim": 128,
  "readiness_class_weight": [
    0.5777286144017337,
    0.7272824755766506,
    1.6949889100216156
  ],
  "readiness_labels": [
    "KEEP",
    "REVIEW",
    "REJECT"
  ],
  "readiness_loss_weight": 1.0,
  "readiness_temperature": 0.5138114435048121,
  "temporal_hidden_size": 256,
  "temporal_intermediate_size": 512,
  "temporal_kernel_size": 3,
  "temporal_num_layers": 3,
  "transformers_version": "5.15.1",
  "vision_config": {
    "_name_or_path": "facebook/dinov2-small",
    "apply_layernorm": true,
    "architectures": [
      "Dinov2Model"
    ],
    "attention_probs_dropout_prob": 0.0,
    "drop_path_rate": 0.0,
    "dtype": "float32",
    "hidden_act": "gelu",
    "hidden_dropout_prob": 0.0,
    "hidden_size": 384,
    "image_size": 518,
    "initializer_range": 0.02,
    "layer_norm_eps": 1e-06,
    "layerscale_value": 1.0,
    "mlp_ratio": 4,
    "model_type": "dinov2",
    "num_attention_heads": 6,
    "num_channels": 3,
    "num_hidden_layers": 12,
    "out_features": [
      "stage12"
    ],
    "out_indices": [
      12
    ],
    "patch_size": 14,
    "qkv_bias": true,
    "reshape_hidden_states": true,
    "stage_names": [
      "stem",
      "stage1",
      "stage2",
      "stage3",
      "stage4",
      "stage5",
      "stage6",
      "stage7",
      "stage8",
      "stage9",
      "stage10",
      "stage11",
      "stage12"
    ],
    "use_mask_token": true,
    "use_swiglu_ffn": false
  }
}