{ "_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": "meeting_room", "1": "cloister", "10": "computerroom", "11": "grocerystore", "12": "hospitalroom", "13": "buffet", "14": "office", "15": "warehouse", "16": "garage", "17": "bookstore", "18": "florist", "19": "locker_room", "2": "stairscase", "20": "inside_bus", "21": "subway", "22": "fastfood_restaurant", "23": "auditorium", "24": "studiomusic", "25": "airport_inside", "26": "pantry", "27": "restaurant_kitchen", "28": "casino", "29": "movietheater", "3": "restaurant", "30": "kitchen", "31": "waitingroom", "32": "artstudio", "33": "toystore", "34": "kindergarden", "35": "trainstation", "36": "bedroom", "37": "mall", "38": "corridor", "39": "bar", "4": "hairsalon", "40": "classroom", "41": "shoeshop", "42": "dentaloffice", "43": "videostore", "44": "laboratorywet", "45": "tv_studio", "46": "church_inside", "47": "operating_room", "48": "jewelleryshop", "49": "bathroom", "5": "children_room", "50": "clothingstore", "51": "closet", "52": "winecellar", "53": "livingroom", "54": "nursery", "55": "gameroom", "56": "inside_subway", "57": "deli", "58": "bakery", "59": "library", "6": "dining_room", "60": "prisoncell", "61": "gym", "62": "concert_hall", "63": "greenhouse", "64": "elevator", "65": "poolinside", "66": "bowling", "7": "lobby", "8": "museum", "9": "laundromat" }, "image_size": 224, "initializer_range": 0.02, "intermediate_size": 3072, "label2id": { "airport_inside": "25", "artstudio": "32", "auditorium": "23", "bakery": "58", "bar": "39", "bathroom": "49", "bedroom": "36", "bookstore": "17", "bowling": "66", "buffet": "13", "casino": "28", "children_room": "5", "church_inside": "46", "classroom": "40", "cloister": "1", "closet": "51", "clothingstore": "50", "computerroom": "10", "concert_hall": "62", "corridor": "38", "deli": "57", "dentaloffice": "42", "dining_room": "6", "elevator": "64", "fastfood_restaurant": "22", "florist": "18", "gameroom": "55", "garage": "16", "greenhouse": "63", "grocerystore": "11", "gym": "61", "hairsalon": "4", "hospitalroom": "12", "inside_bus": "20", "inside_subway": "56", "jewelleryshop": "48", "kindergarden": "34", "kitchen": "30", "laboratorywet": "44", "laundromat": "9", "library": "59", "livingroom": "53", "lobby": "7", "locker_room": "19", "mall": "37", "meeting_room": "0", "movietheater": "29", "museum": "8", "nursery": "54", "office": "14", "operating_room": "47", "pantry": "26", "poolinside": "65", "prisoncell": "60", "restaurant": "3", "restaurant_kitchen": "27", "shoeshop": "41", "stairscase": "2", "studiomusic": "24", "subway": "21", "toystore": "33", "trainstation": "35", "tv_studio": "45", "videostore": "43", "waitingroom": "31", "warehouse": "15", "winecellar": "52" }, "layer_norm_eps": 1e-12, "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": "4.33.3" }