{
  "_commit_hash": null,
  "architectures": [
    "DPTForDepthEstimation"
  ],
  "attention_probs_dropout_prob": 0.1,
  "auxiliary_loss_weight": 0.4,
  "backbone_config": null,
  "backbone_featmap_shape": null,
  "backbone_out_indices": [
    0,
    1,
    2,
    3
  ],
  "fusion_hidden_size": 256,
  "head_in_index": -1,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 32,
  "image_size": 32,
  "initializer_range": 0.02,
  "intermediate_size": 37,
  "is_hybrid": false,
  "layer_norm_eps": 1e-12,
  "model_type": "dpt",
  "neck_hidden_sizes": [
    96,
    192,
    384,
    768
  ],
  "neck_ignore_stages": [],
  "num_attention_heads": 4,
  "num_channels": 3,
  "num_hidden_layers": 4,
  "patch_size": 16,
  "qkv_bias": true,
  "readout_type": "project",
  "reassemble_factors": [
    4,
    2,
    1,
    0.5
  ],
  "semantic_classifier_dropout": 0.1,
  "semantic_loss_ignore_index": 255,
  "torch_dtype": "float32",
  "transformers_version": null,
  "use_auxiliary_head": true,
  "use_batch_norm_in_fusion_residual": false
}