config.json

{
  "architecture": "vit_base_patch16_224",
  "num_classes": 0,
  "num_features": 128,
  "global_pool": "token",
  "model_args": {
    "img_size": 56,
    "patch_size": 4,
    "in_chans": 5,
    "embed_dim": 128,
    "depth": 12,
    "num_heads": 8,
    "mlp_ratio": 4,
    "reg_tokens": 8,
    "init_values": 1e-05,
    "dynamic_img_size": true,
    "num_classes": 0
  },
  "pretrained_cfg": {
    "custom_load": false,
    "input_size": [
      5,
      56,
      56
    ],
    "fixed_input_size": true,
    "interpolation": "bilinear",
    "crop_pct": 1.0,
    "crop_mode": "center",
    "mean": [
      0.0972,
      0.7408,
      0.103,
      0.0761,
      0.0893
    ],
    "std": [
      0.141,
      0.118,
      0.1694,
      0.1051,
      0.0972
    ],
    "num_classes": 0,
    "pool_size": null,
    "first_conv": null,
    "classifier": "head"
  },
  "scdino_variant": "dinov2",
  "scdino_preprocessing": {
    "norm_type": "clip_max",
    "clip_mode": "clip_and_scale",
    "input_layout": "HWC",
    "max_vals_clip": [
      3713.48,
      1297.59,
      6904.38,
      3412.43,
      3133.96
    ],
    "mean": [
      0.0972,
      0.7408,
      0.103,
      0.0761,
      0.0893
    ],
    "std": [
      0.141,
      0.118,
      0.1694,
      0.1051,
      0.0972
    ],
    "resize": 56
  }
}