{
  "_name_or_path": "google/vit-base-patch16-224-in21k",
  "architectures": [
    "ViTForImageClassification"
  ],
  "attention_probs_dropout_prob": 0.0,
  "encoder_stride": 16,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.0,
  "hidden_size": 768,
  "id2label": {
    "0": "Batray",
    "1": "Bird",
    "2": "Boat",
    "3": "Buoy",
    "4": "Egregia",
    "5": "Foam",
    "6": "Jelly",
    "7": "Kelp",
    "8": "Mola",
    "9": "Mooring",
    "10": "Otter",
    "11": "Person",
    "12": "Pinniped",
    "13": "Poop",
    "14": "RIB",
    "15": "Reflectance",
    "16": "Secci_Disc",
    "17": "Shark",
    "18": "Surfboard",
    "19": "Wave",
    "20": "Whale",
    "21": "Wood"
  },
  "image_size": 224,
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "Batray": 0,
    "Bird": 1,
    "Boat": 2,
    "Buoy": 3,
    "Egregia": 4,
    "Foam": 5,
    "Jelly": 6,
    "Kelp": 7,
    "Mola": 8,
    "Mooring": 9,
    "Otter": 10,
    "Person": 11,
    "Pinniped": 12,
    "Poop": 13,
    "RIB": 14,
    "Reflectance": 15,
    "Secci_Disc": 16,
    "Shark": 17,
    "Surfboard": 18,
    "Wave": 19,
    "Whale": 20,
    "Wood": 21
  },
  "layer_norm_eps": 1e-12,
  "model_type": "vit",
  "num_attention_heads": 12,
  "num_channels": 3,
  "num_hidden_layers": 12,
  "patch_size": 16,
  "qkv_bias": true,
  "torch_dtype": "float32",
  "transformers_version": "4.46.3"
}