config.json

{
  "architectures": [
    "SMSelectiveViTModelForClassification"
  ],
  "atten_dim": 768,
  "attention_scale": 8.0,
  "auto_map": {
    "AutoConfig": "configuration_selectivevit.SMSelectiveViTConfig",
    "AutoModelForImageClassification": "modeling_selectivevit.SMSelectiveViTModelForClassification"
  },
  "channels": 3,
  "depth": 12,
  "drop_path": 0.2,
  "dropout": 0.0,
  "dtype": "float32",
  "embed_dim": 768,
  "image_size": 224,
  "mask_threshold": 0.05,
  "mlp_dim": 3072,
  "model_type": "softmasked_selective_vit",
  "num_classes": 1000,
  "num_heads": 12,
  "patch_drop": 0.05,
  "patch_size": 16,
  "transformers_version": "4.57.3",
  "use_distil_token": true
}