{ "_name_or_path": "apple/mobilevit-x-small", "architectures": [ "MobileViTForImageClassification" ], "aspp_dropout_prob": 0.1, "aspp_out_channels": 256, "atrous_rates": [ 6, 12, 18 ], "attention_probs_dropout_prob": 0.0, "classifier_dropout_prob": 0.1, "conv_kernel_size": 3, "expand_ratio": 4.0, "hidden_act": "silu", "hidden_dropout_prob": 0.1, "hidden_sizes": [ 96, 120, 144 ], "id2label": { "0": "aircraft carrier", "1": "airplane", "10": "asparagus", "100": "dumbbell", "101": "ear", "102": "elbow", "103": "elephant", "104": "envelope", "105": "eraser", "106": "eye", "107": "eyeglasses", "108": "face", "109": "fan", "11": "axe", "110": "feather", "111": "fence", "112": "finger", "113": "fire hydrant", "114": "fireplace", "115": "firetruck", "116": "fish", "117": "flamingo", "118": "flashlight", "119": "flip flops", "12": "backpack", "120": "floor lamp", "121": "flower", "122": "flying saucer", "123": "foot", "124": "fork", "125": "frog", "126": "frying pan", "127": "garden hose", "128": "garden", "129": "giraffe", "13": "banana", "130": "goatee", "131": "golf club", "132": "grapes", "133": "grass", "134": "guitar", "135": "hamburger", "136": "hammer", "137": "hand", "138": "harp", "139": "hat", "14": "bandage", "140": "headphones", "141": "hedgehog", "142": "helicopter", "143": "helmet", "144": "hexagon", "145": "hockey puck", "146": "hockey stick", "147": "horse", "148": "hospital", "149": "hot air balloon", "15": "barn", "150": "hot dog", "151": "hot tub", "152": "hourglass", "153": "house plant", "154": "house", "155": "hurricane", "156": "ice cream", "157": "jacket", "158": "jail", "159": "kangaroo", "16": "baseball bat", "160": "key", "161": "keyboard", "162": "knee", "163": "knife", "164": "ladder", "165": "lantern", "166": "laptop", "167": "leaf", "168": "leg", "169": "light bulb", "17": "baseball", "170": "lighter", "171": "lighthouse", "172": "lightning", "173": "line", "174": "lion", "175": "lipstick", "176": "lobster", "177": "lollipop", "178": "mailbox", "179": "map", "18": "basket", "180": "marker", "181": "matches", "182": "megaphone", "183": "mermaid", "184": "microphone", "185": "microwave", "186": "monkey", "187": "moon", "188": "mosquito", "189": "motorbike", "19": "basketball", "190": "mountain", "191": "mouse", "192": "moustache", "193": "mouth", "194": "mug", "195": "mushroom", "196": "nail", "197": "necklace", "198": "nose", "199": "ocean", "2": "alarm clock", "20": "bat", "200": "octagon", "201": "octopus", "202": "onion", "203": "oven", "204": "owl", "205": "paint can", "206": "paintbrush", "207": "palm tree", "208": "panda", "209": "pants", "21": "bathtub", "210": "paper clip", "211": "parachute", "212": "parrot", "213": "passport", "214": "peanut", "215": "pear", "216": "peas", "217": "pencil", "218": "penguin", "219": "piano", "22": "beach", "220": "pickup truck", "221": "picture frame", "222": "pig", "223": "pillow", "224": "pineapple", "225": "pizza", "226": "pliers", "227": "police car", "228": "pond", "229": "pool", "23": "bear", "230": "popsicle", "231": "postcard", "232": "potato", "233": "power outlet", "234": "purse", "235": "rabbit", "236": "raccoon", "237": "radio", "238": "rain", "239": "rainbow", "24": "beard", "240": "rake", "241": "remote control", "242": "rhinoceros", "243": "rifle", "244": "river", "245": "roller coaster", "246": "rollerskates", "247": "sailboat", "248": "sandwich", "249": "saw", "25": "bed", "250": "saxophone", "251": "school bus", "252": "scissors", "253": "scorpion", "254": "screwdriver", "255": "sea turtle", "256": "see saw", "257": "shark", "258": "sheep", "259": "shoe", "26": "bee", "260": "shorts", "261": "shovel", "262": "sink", "263": "skateboard", "264": "skull", "265": "skyscraper", "266": "sleeping bag", "267": "smiley face", "268": "snail", "269": "snake", "27": "belt", "270": "snorkel", "271": "snowflake", "272": "snowman", "273": "soccer ball", "274": "sock", "275": "speedboat", "276": "spider", "277": "spoon", "278": "spreadsheet", "279": "square", "28": "bench", "280": "squiggle", "281": "squirrel", "282": "stairs", "283": "star", "284": "steak", "285": "stereo", "286": "stethoscope", "287": "stitches", "288": "stop sign", "289": "stove", "29": "bicycle", "290": "strawberry", "291": "streetlight", "292": "string bean", "293": "submarine", "294": "suitcase", "295": "sun", "296": "swan", "297": "sweater", "298": "swing set", "299": "sword", "3": "ambulance", "30": "binoculars", "300": "syringe", "301": "t-shirt", "302": "table", "303": "teapot", "304": "teddy-bear", "305": "telephone", "306": "television", "307": "tennis racquet", "308": "tent", "309": "The Eiffel Tower", "31": "bird", "310": "The Great Wall of China", "311": "The Mona Lisa", "312": "tiger", "313": "toaster", "314": "toe", "315": "toilet", "316": "tooth", "317": "toothbrush", "318": "toothpaste", "319": "tornado", "32": "birthday cake", "320": "tractor", "321": "traffic light", "322": "train", "323": "tree", "324": "triangle", "325": "trombone", "326": "truck", "327": "trumpet", "328": "umbrella", "329": "underwear", "33": "blackberry", "330": "van", "331": "vase", "332": "violin", "333": "washing machine", "334": "watermelon", "335": "waterslide", "336": "whale", "337": "wheel", "338": "windmill", "339": "wine bottle", "34": "blueberry", "340": "wine glass", "341": "wristwatch", "342": "yoga", "343": "zebra", "344": "zigzag", "35": "book", "36": "boomerang", "37": "bottlecap", "38": "bowtie", "39": "bracelet", "4": "angel", "40": "brain", "41": "bread", "42": "bridge", "43": "broccoli", "44": "broom", "45": "bucket", "46": "bulldozer", "47": "bus", "48": "bush", "49": "butterfly", "5": "animal migration", "50": "cactus", "51": "cake", "52": "calculator", "53": "calendar", "54": "camel", "55": "camera", "56": "camouflage", "57": "campfire", "58": "candle", "59": "cannon", "6": "ant", "60": "canoe", "61": "car", "62": "carrot", "63": "castle", "64": "cat", "65": "ceiling fan", "66": "cell phone", "67": "cello", "68": "chair", "69": "chandelier", "7": "anvil", "70": "church", "71": "circle", "72": "clarinet", "73": "clock", "74": "cloud", "75": "coffee cup", "76": "compass", "77": "computer", "78": "cookie", "79": "cooler", "8": "apple", "80": "couch", "81": "cow", "82": "crab", "83": "crayon", "84": "crocodile", "85": "crown", "86": "cruise ship", "87": "cup", "88": "diamond", "89": "dishwasher", "9": "arm", "90": "diving board", "91": "dog", "92": "dolphin", "93": "donut", "94": "door", "95": "dragon", "96": "dresser", "97": "drill", "98": "drums", "99": "duck" }, "image_size": 28, "initializer_range": 0.02, "label2id": { "The Eiffel Tower": "309", "The Great Wall of China": "310", "The Mona Lisa": "311", "aircraft carrier": "0", "airplane": "1", "alarm clock": "2", "ambulance": "3", "angel": "4", "animal migration": "5", "ant": "6", "anvil": "7", "apple": "8", "arm": "9", "asparagus": "10", "axe": "11", "backpack": "12", "banana": "13", "bandage": "14", "barn": "15", "baseball": "17", "baseball bat": "16", "basket": "18", "basketball": "19", "bat": "20", "bathtub": "21", "beach": "22", "bear": "23", "beard": "24", "bed": "25", "bee": "26", "belt": "27", "bench": "28", "bicycle": "29", "binoculars": "30", "bird": "31", "birthday cake": "32", "blackberry": "33", "blueberry": "34", "book": "35", "boomerang": "36", "bottlecap": "37", "bowtie": "38", "bracelet": "39", "brain": "40", "bread": "41", "bridge": "42", "broccoli": "43", "broom": "44", "bucket": "45", "bulldozer": "46", "bus": "47", "bush": "48", "butterfly": "49", "cactus": "50", "cake": "51", "calculator": "52", "calendar": "53", "camel": "54", "camera": "55", "camouflage": "56", "campfire": "57", "candle": "58", "cannon": "59", "canoe": "60", "car": "61", "carrot": "62", "castle": "63", "cat": "64", "ceiling fan": "65", "cell phone": "66", "cello": "67", "chair": "68", "chandelier": "69", "church": "70", "circle": "71", "clarinet": "72", "clock": "73", "cloud": "74", "coffee cup": "75", "compass": "76", "computer": "77", "cookie": "78", "cooler": "79", "couch": "80", "cow": "81", "crab": "82", "crayon": "83", "crocodile": "84", "crown": "85", "cruise ship": "86", "cup": "87", "diamond": "88", "dishwasher": "89", "diving board": "90", "dog": "91", "dolphin": "92", "donut": "93", "door": "94", "dragon": "95", "dresser": "96", "drill": "97", "drums": "98", "duck": "99", "dumbbell": "100", "ear": "101", "elbow": "102", "elephant": "103", "envelope": "104", "eraser": "105", "eye": "106", "eyeglasses": "107", "face": "108", "fan": "109", "feather": "110", "fence": "111", "finger": "112", "fire hydrant": "113", "fireplace": "114", "firetruck": "115", "fish": "116", "flamingo": "117", "flashlight": "118", "flip flops": "119", "floor lamp": "120", "flower": "121", "flying saucer": "122", "foot": "123", "fork": "124", "frog": "125", "frying pan": "126", "garden": "128", "garden hose": "127", "giraffe": "129", "goatee": "130", "golf club": "131", "grapes": "132", "grass": "133", "guitar": "134", "hamburger": "135", "hammer": "136", "hand": "137", "harp": "138", "hat": "139", "headphones": "140", "hedgehog": "141", "helicopter": "142", "helmet": "143", "hexagon": "144", "hockey puck": "145", "hockey stick": "146", "horse": "147", "hospital": "148", "hot air balloon": "149", "hot dog": "150", "hot tub": "151", "hourglass": "152", "house": "154", "house plant": "153", "hurricane": "155", "ice cream": "156", "jacket": "157", "jail": "158", "kangaroo": "159", "key": "160", "keyboard": "161", "knee": "162", "knife": "163", "ladder": "164", "lantern": "165", "laptop": "166", "leaf": "167", "leg": "168", "light bulb": "169", "lighter": "170", "lighthouse": "171", "lightning": "172", "line": "173", "lion": "174", "lipstick": "175", "lobster": "176", "lollipop": "177", "mailbox": "178", "map": "179", "marker": "180", "matches": "181", "megaphone": "182", "mermaid": "183", "microphone": "184", "microwave": "185", "monkey": "186", "moon": "187", "mosquito": "188", "motorbike": "189", "mountain": "190", "mouse": "191", "moustache": "192", "mouth": "193", "mug": "194", "mushroom": "195", "nail": "196", "necklace": "197", "nose": "198", "ocean": "199", "octagon": "200", "octopus": "201", "onion": "202", "oven": "203", "owl": "204", "paint can": "205", "paintbrush": "206", "palm tree": "207", "panda": "208", "pants": "209", "paper clip": "210", "parachute": "211", "parrot": "212", "passport": "213", "peanut": "214", "pear": "215", "peas": "216", "pencil": "217", "penguin": "218", "piano": "219", "pickup truck": "220", "picture frame": "221", "pig": "222", "pillow": "223", "pineapple": "224", "pizza": "225", "pliers": "226", "police car": "227", "pond": "228", "pool": "229", "popsicle": "230", "postcard": "231", "potato": "232", "power outlet": "233", "purse": "234", "rabbit": "235", "raccoon": "236", "radio": "237", "rain": "238", "rainbow": "239", "rake": "240", "remote control": "241", "rhinoceros": "242", "rifle": "243", "river": "244", "roller coaster": "245", "rollerskates": "246", "sailboat": "247", "sandwich": "248", "saw": "249", "saxophone": "250", "school bus": "251", "scissors": "252", "scorpion": "253", "screwdriver": "254", "sea turtle": "255", "see saw": "256", "shark": "257", "sheep": "258", "shoe": "259", "shorts": "260", "shovel": "261", "sink": "262", "skateboard": "263", "skull": "264", "skyscraper": "265", "sleeping bag": "266", "smiley face": "267", "snail": "268", "snake": "269", "snorkel": "270", "snowflake": "271", "snowman": "272", "soccer ball": "273", "sock": "274", "speedboat": "275", "spider": "276", "spoon": "277", "spreadsheet": "278", "square": "279", "squiggle": "280", "squirrel": "281", "stairs": "282", "star": "283", "steak": "284", "stereo": "285", "stethoscope": "286", "stitches": "287", "stop sign": "288", "stove": "289", "strawberry": "290", "streetlight": "291", "string bean": "292", "submarine": "293", "suitcase": "294", "sun": "295", "swan": "296", "sweater": "297", "swing set": "298", "sword": "299", "syringe": "300", "t-shirt": "301", "table": "302", "teapot": "303", "teddy-bear": "304", "telephone": "305", "television": "306", "tennis racquet": "307", "tent": "308", "tiger": "312", "toaster": "313", "toe": "314", "toilet": "315", "tooth": "316", "toothbrush": "317", "toothpaste": "318", "tornado": "319", "tractor": "320", "traffic light": "321", "train": "322", "tree": "323", "triangle": "324", "trombone": "325", "truck": "326", "trumpet": "327", "umbrella": "328", "underwear": "329", "van": "330", "vase": "331", "violin": "332", "washing machine": "333", "watermelon": "334", "waterslide": "335", "whale": "336", "wheel": "337", "windmill": "338", "wine bottle": "339", "wine glass": "340", "wristwatch": "341", "yoga": "342", "zebra": "343", "zigzag": "344" }, "layer_norm_eps": 1e-05, "mlp_ratio": 2.0, "model_type": "mobilevit", "neck_hidden_sizes": [ 16, 32, 48, 64, 80, 96, 384 ], "num_attention_heads": 4, "num_channels": 1, "output_stride": 32, "patch_size": 2, "problem_type": "single_label_classification", "qkv_bias": true, "semantic_loss_ignore_index": 255, "torch_dtype": "float32", "transformers_version": "4.42.4" }