config.json

{
  "_name_or_path": "dandelin/vilt-b32-mlm",
  "architectures": [
    "ViltForQuestionAnswering"
  ],
  "attention_probs_dropout_prob": 0.0,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.0,
  "hidden_size": 768,
  "id2label": {
    "0": "room",
    "1": "united states",
    "2": "airport",
    "3": "australia",
    "4": "usa",
    "5": "windows",
    "6": "home",
    "7": "fridge",
    "8": "in car",
    "9": "park",
    "10": "england",
    "11": "on sidewalk",
    "12": "nothing",
    "13": "mouse",
    "14": "sun",
    "15": "smile",
    "16": "refrigerator",
    "17": "british",
    "18": "freezer",
    "19": "hotel",
    "20": "car",
    "21": "indoors",
    "22": "woods",
    "23": "buildings",
    "24": "uk",
    "25": "nowhere",
    "26": "cows",
    "27": "dog",
    "28": "bus",
    "29": "america",
    "30": "living room",
    "31": "ground",
    "32": "us",
    "33": "inside",
    "34": "yes",
    "35": "hallway"
  },
  "image_size": 384,
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "airport": 2,
    "america": 29,
    "australia": 3,
    "british": 17,
    "buildings": 23,
    "bus": 28,
    "car": 20,
    "cows": 26,
    "dog": 27,
    "england": 10,
    "freezer": 18,
    "fridge": 7,
    "ground": 31,
    "hallway": 35,
    "home": 6,
    "hotel": 19,
    "in car": 8,
    "indoors": 21,
    "inside": 33,
    "living room": 30,
    "mouse": 13,
    "nothing": 12,
    "nowhere": 25,
    "on sidewalk": 11,
    "park": 9,
    "refrigerator": 16,
    "room": 0,
    "smile": 15,
    "sun": 14,
    "uk": 24,
    "united states": 1,
    "us": 32,
    "usa": 4,
    "windows": 5,
    "woods": 22,
    "yes": 34
  },
  "layer_norm_eps": 1e-12,
  "max_image_length": -1,
  "max_position_embeddings": 40,
  "modality_type_vocab_size": 2,
  "model_type": "vilt",
  "num_attention_heads": 12,
  "num_channels": 3,
  "num_hidden_layers": 12,
  "num_images": -1,
  "patch_size": 32,
  "qkv_bias": true,
  "tie_word_embeddings": false,
  "torch_dtype": "float32",
  "transformers_version": "4.40.2",
  "type_vocab_size": 2,
  "vocab_size": 30522
}