image_encoder/config.json

{
  "apply_layernorm": true,
  "architectures": [
    "Dinov2WithRegistersModel"
  ],
  "attention_probs_dropout_prob": 0.0,
  "drop_path_rate": 0.0,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.0,
  "hidden_size": 1024,
  "image_size": 518,
  "initializer_range": 0.02,
  "interpolate_antialias": true,
  "interpolate_offset": 0.0,
  "layer_norm_eps": 1e-06,
  "layerscale_value": 1.0,
  "mlp_ratio": 4,
  "model_type": "dinov2_with_registers",
  "num_attention_heads": 16,
  "num_channels": 3,
  "num_hidden_layers": 24,
  "num_register_tokens": 4,
  "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",
    "stage13",
    "stage14",
    "stage15",
    "stage16",
    "stage17",
    "stage18",
    "stage19",
    "stage20",
    "stage21",
    "stage22",
    "stage23",
    "stage24"
  ],
  "torch_dtype": "float32",
  "transformers_version": "4.55.4",
  "use_swiglu_ffn": false
}