config.json

{
  "act_layer": null,
  "attn_drop_rate": 0.0,
  "backbone": "hybrid",
  "channel_dims": [
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    528,
    768,
    768,
    768
  ],
  "channels": 256,
  "cls_attn_layers": 2,
  "decoder_dropout": 0.1,
  "decoder_hidden_size": 768,
  "depths": [
    3,
    7
  ],
  "dims": [
    128,
    256,
    512,
    1024
  ],
  "drop_path_rate": 0.0,
  "drop_rate": 0.0,
  "embed_dim": 528,
  "eta": 1.0,
  "feat_downsample": false,
  "feature_strides": [
    4,
    8,
    16,
    32
  ],
  "hybrid_patch_size": 2,
  "img_size": [
    224,
    224
  ],
  "initializer_range": 1.0,
  "mlp_ratio": 4.0,
  "model_type": "fan",
  "norm_layer": null,
  "num_attention_heads": [
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    11,
    16,
    16,
    16
  ],
  "num_channels": 3,
  "num_classes": 1000,
  "num_hidden_layers": 23,
  "out_index": -1,
  "patch_size": 16,
  "qkv_bias": true,
  "reshape_last_stage": false,
  "rounding_mode": "floor",
  "se_mlp": false,
  "segmentation_in_channels": [
    128,
    256,
    480,
    480
  ],
  "semantic_loss_ignore_index": -100,
  "sharpen_attn": false,
  "tokens_norm": true,
  "transformers_version": "4.24.0.dev0",
  "use_checkpoint": false,
  "use_head": false,
  "use_pos_embed": true
}