{
  "architectures": [
    "DepthAnythingForDepthEstimation"
  ],
  "backbone_config": {
    "apply_layernorm": true,
    "attention_probs_dropout_prob": 0.0,
    "drop_path_rate": 0.0,
    "hidden_act": "gelu",
    "hidden_dropout_prob": 0.0,
    "hidden_size": 4,
    "image_size": 32,
    "initializer_range": 0.02,
    "intermediate_size": 8,
    "is_training": true,
    "layer_norm_eps": 1e-06,
    "layerscale_value": 1.0,
    "mlp_ratio": 4,
    "model_type": "dinov2",
    "num_attention_heads": 2,
    "num_channels": 3,
    "num_hidden_layers": 2,
    "out_features": [
      "stage1",
      "stage2"
    ],
    "out_indices": [
      1,
      2
    ],
    "patch_size": 16,
    "qkv_bias": true,
    "reshape_hidden_states": false,
    "stage_names": [
      "stem",
      "stage1",
      "stage2"
    ],
    "use_mask_token": true,
    "use_swiglu_ffn": false
  },
  "depth_estimation_type": "relative",
  "dtype": "float32",
  "fusion_hidden_size": 6,
  "head_hidden_size": 32,
  "head_in_index": -1,
  "initializer_range": 0.02,
  "max_depth": 1,
  "model_type": "depth_anything",
  "neck_hidden_sizes": [
    2,
    2
  ],
  "patch_size": 16,
  "reassemble_factors": [
    4,
    2,
    1,
    0.5
  ],
  "reassemble_hidden_size": 4,
  "transformers_version": "5.16.0.dev0"
}