config.json

{
  "_name_or_path": "microsoft/cvt-w24-384-22k",
  "architectures": [
    "CvtForImageClassification"
  ],
  "attention_drop_rate": [
    0.0,
    0.0,
    0.0
  ],
  "cls_token": [
    false,
    false,
    true
  ],
  "depth": [
    2,
    2,
    20
  ],
  "drop_path_rate": [
    0.0,
    0.0,
    0.1
  ],
  "drop_rate": [
    0.0,
    0.0,
    0.0
  ],
  "embed_dim": [
    192,
    768,
    1024
  ],
  "id2label": {
    "0": "leaflet",
    "1": "pinnate"
  },
  "initializer_range": 0.02,
  "kernel_qkv": [
    3,
    3,
    3
  ],
  "label2id": {
    "leaflet": 0,
    "pinnate": 1
  },
  "layer_norm_eps": 1e-12,
  "mlp_ratio": [
    4.0,
    4.0,
    4.0
  ],
  "model_type": "cvt",
  "num_channels": 3,
  "num_heads": [
    3,
    12,
    16
  ],
  "padding_kv": [
    1,
    1,
    1
  ],
  "padding_q": [
    1,
    1,
    1
  ],
  "patch_padding": [
    2,
    1,
    1
  ],
  "patch_sizes": [
    7,
    3,
    3
  ],
  "patch_stride": [
    4,
    2,
    2
  ],
  "problem_type": "single_label_classification",
  "qkv_bias": [
    true,
    true,
    true
  ],
  "qkv_projection_method": [
    "dw_bn",
    "dw_bn",
    "dw_bn"
  ],
  "stride_kv": [
    2,
    2,
    2
  ],
  "stride_q": [
    1,
    1,
    1
  ],
  "torch_dtype": "float32",
  "transformers_version": "4.38.1"
}