{
  "_name_or_path": "microsoft/resnet-50",
  "architectures": [
    "ViTForImageClassification"
  ],
  "attention_probs_dropout_prob": 0.0,
  "depths": [
    3,
    4,
    6,
    3
  ],
  "downsample_in_first_stage": false,
  "embedding_size": 64,
  "encoder_stride": 16,
  "hidden_act": "relu",
  "hidden_dropout_prob": 0.0,
  "hidden_size": 768,
  "hidden_sizes": [
    256,
    512,
    1024,
    2048
  ],
  "id2label": {
    "0": "Cloudy",
    "1": "Normal",
    "2": "Rainy",
    "3": "Snowy"
  },
  "image_size": 224,
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "Cloudy": 0,
    "Normal": 1,
    "Rainy": 2,
    "Snowy": 3
  },
  "layer_norm_eps": 1e-12,
  "layer_type": "bottleneck",
  "model_type": "vit",
  "num_attention_heads": 12,
  "num_channels": 3,
  "num_hidden_layers": 12,
  "patch_size": 16,
  "problem_type": "single_label_classification",
  "qkv_bias": true,
  "torch_dtype": "float32",
  "transformers_version": "4.30.2"
}