{ "_commit_hash": null, "architectures": [ "VisionEncoderDecoderModel" ], "decoder": { "_name_or_path": "bert-base-uncased", "add_cross_attention": true, "architectures": [ "BertForMaskedLM" ], "attention_probs_dropout_prob": 0.1, "bad_words_ids": null, "begin_suppress_tokens": null, "bos_token_id": null, "chunk_size_feed_forward": 0, "classifier_dropout": null, "cross_attention_hidden_size": null, "decoder_start_token_id": null, "diversity_penalty": 0.0, "do_sample": false, "early_stopping": false, "encoder_no_repeat_ngram_size": 0, "eos_token_id": null, "exponential_decay_length_penalty": null, "finetuning_task": null, "forced_bos_token_id": null, "forced_eos_token_id": null, "gradient_checkpointing": false, "hidden_act": "gelu", "hidden_dropout_prob": 0.1, "hidden_size": 768, "id2label": { "0": "LABEL_0", "1": "LABEL_1" }, "initializer_range": 0.02, "intermediate_size": 3072, "is_decoder": true, "is_encoder_decoder": false, "label2id": { "LABEL_0": 0, "LABEL_1": 1 }, "layer_norm_eps": 1e-12, "length_penalty": 1.0, "max_length": 20, "max_position_embeddings": 512, "min_length": 0, "model_type": "bert", "no_repeat_ngram_size": 0, "num_attention_heads": 12, "num_beam_groups": 1, "num_beams": 1, "num_hidden_layers": 12, "num_return_sequences": 1, "output_attentions": false, "output_hidden_states": false, "output_scores": false, "pad_token_id": 0, "position_embedding_type": "absolute", "prefix": null, "problem_type": null, "pruned_heads": {}, "remove_invalid_values": false, "repetition_penalty": 1.0, "return_dict": true, "return_dict_in_generate": false, "sep_token_id": null, "suppress_tokens": null, "task_specific_params": null, "temperature": 1.0, "tf_legacy_loss": false, "tie_encoder_decoder": false, "tie_word_embeddings": true, "tokenizer_class": null, "top_k": 50, "top_p": 1.0, "torch_dtype": null, "torchscript": false, "transformers_version": "4.30.2", "type_vocab_size": 2, "typical_p": 1.0, "use_bfloat16": false, "use_cache": true, "vocab_size": 30522 }, "decoder_start_token_id": 101, "encoder": { "_name_or_path": "facebook/timesformer-base-finetuned-k400", "add_cross_attention": false, "architectures": [ "TimesformerForVideoClassification" ], "attention_probs_dropout_prob": 0.0, "attention_type": "divided_space_time", "bad_words_ids": null, "begin_suppress_tokens": null, "bos_token_id": null, "chunk_size_feed_forward": 0, "cross_attention_hidden_size": null, "decoder_start_token_id": null, "diversity_penalty": 0.0, "do_sample": false, "drop_path_rate": 0, "early_stopping": false, "encoder_no_repeat_ngram_size": 0, "eos_token_id": null, "exponential_decay_length_penalty": null, "finetuning_task": null, "forced_bos_token_id": null, "forced_eos_token_id": null, "hidden_act": "gelu", "hidden_dropout_prob": 0.0, "hidden_size": 768, "id2label": { "0": "abseiling", "1": "air drumming", "2": "answering questions", "3": "applauding", "4": "applying cream", "5": "archery", "6": "arm wrestling", "7": "arranging flowers", "8": "assembling computer", "9": "auctioning", "10": "baby waking up", "11": "baking cookies", "12": "balloon blowing", "13": "bandaging", "14": "barbequing", "15": "bartending", "16": "beatboxing", "17": "bee keeping", "18": "belly dancing", "19": "bench pressing", "20": "bending back", "21": "bending metal", "22": "biking through snow", "23": "blasting sand", "24": "blowing glass", "25": "blowing leaves", "26": "blowing nose", "27": "blowing out candles", "28": "bobsledding", "29": "bookbinding", "30": "bouncing on trampoline", "31": "bowling", "32": "braiding hair", "33": "breading or breadcrumbing", "34": "breakdancing", "35": "brush painting", "36": "brushing hair", "37": "brushing teeth", "38": "building cabinet", "39": "building shed", "40": "bungee jumping", "41": "busking", "42": "canoeing or kayaking", "43": "capoeira", "44": "carrying baby", "45": "cartwheeling", "46": "carving pumpkin", "47": "catching fish", "48": "catching or throwing baseball", "49": "catching or throwing frisbee", "50": "catching or throwing softball", "51": "celebrating", "52": "changing oil", "53": "changing wheel", "54": "checking tires", "55": "cheerleading", "56": "chopping wood", "57": "clapping", "58": "clay pottery making", "59": "clean and jerk", "60": "cleaning floor", "61": "cleaning gutters", "62": "cleaning pool", "63": "cleaning shoes", "64": "cleaning toilet", "65": "cleaning windows", "66": "climbing a rope", "67": "climbing ladder", "68": "climbing tree", "69": "contact juggling", "70": "cooking chicken", "71": "cooking egg", "72": "cooking on campfire", "73": "cooking sausages", "74": "counting money", "75": "country line dancing", "76": "cracking neck", "77": "crawling baby", "78": "crossing river", "79": "crying", "80": "curling hair", "81": "cutting nails", "82": "cutting pineapple", "83": "cutting watermelon", "84": "dancing ballet", "85": "dancing charleston", "86": "dancing gangnam style", "87": "dancing macarena", "88": "deadlifting", "89": "decorating the christmas tree", "90": "digging", "91": "dining", "92": "disc golfing", "93": "diving cliff", "94": "dodgeball", "95": "doing aerobics", "96": "doing laundry", "97": "doing nails", "98": "drawing", "99": "dribbling basketball", "100": "drinking", "101": "drinking beer", "102": "drinking shots", "103": "driving car", "104": "driving tractor", "105": "drop kicking", "106": "drumming fingers", "107": "dunking basketball", "108": "dying hair", "109": "eating burger", "110": "eating cake", "111": "eating carrots", "112": "eating chips", "113": "eating doughnuts", "114": "eating hotdog", "115": "eating ice cream", "116": "eating spaghetti", "117": "eating watermelon", "118": "egg hunting", "119": "exercising arm", "120": "exercising with an exercise ball", "121": "extinguishing fire", "122": "faceplanting", "123": "feeding birds", "124": "feeding fish", "125": "feeding goats", "126": "filling eyebrows", "127": "finger snapping", "128": "fixing hair", "129": "flipping pancake", "130": "flying kite", "131": "folding clothes", "132": "folding napkins", "133": "folding paper", "134": "front raises", "135": "frying vegetables", "136": "garbage collecting", "137": "gargling", "138": "getting a haircut", "139": "getting a tattoo", "140": "giving or receiving award", "141": "golf chipping", "142": "golf driving", "143": "golf putting", "144": "grinding meat", "145": "grooming dog", "146": "grooming horse", "147": "gymnastics tumbling", "148": "hammer throw", "149": "headbanging", "150": "headbutting", "151": "high jump", "152": "high kick", "153": "hitting baseball", "154": "hockey stop", "155": "holding snake", "156": "hopscotch", "157": "hoverboarding", "158": "hugging", "159": "hula hooping", "160": "hurdling", "161": "hurling (sport)", "162": "ice climbing", "163": "ice fishing", "164": "ice skating", "165": "ironing", "166": "javelin throw", "167": "jetskiing", "168": "jogging", "169": "juggling balls", "170": "juggling fire", "171": "juggling soccer ball", "172": "jumping into pool", "173": "jumpstyle dancing", "174": "kicking field goal", "175": "kicking soccer ball", "176": "kissing", "177": "kitesurfing", "178": "knitting", "179": "krumping", "180": "laughing", "181": "laying bricks", "182": "long jump", "183": "lunge", "184": "making a cake", "185": "making a sandwich", "186": "making bed", "187": "making jewelry", "188": "making pizza", "189": "making snowman", "190": "making sushi", "191": "making tea", "192": "marching", "193": "massaging back", "194": "massaging feet", "195": "massaging legs", "196": "massaging person's head", "197": "milking cow", "198": "mopping floor", "199": "motorcycling", "200": "moving furniture", "201": "mowing lawn", "202": "news anchoring", "203": "opening bottle", "204": "opening present", "205": "paragliding", "206": "parasailing", "207": "parkour", "208": "passing American football (in game)", "209": "passing American football (not in game)", "210": "peeling apples", "211": "peeling potatoes", "212": "petting animal (not cat)", "213": "petting cat", "214": "picking fruit", "215": "planting trees", "216": "plastering", "217": "playing accordion", "218": "playing badminton", "219": "playing bagpipes", "220": "playing basketball", "221": "playing bass guitar", "222": "playing cards", "223": "playing cello", "224": "playing chess", "225": "playing clarinet", "226": "playing controller", "227": "playing cricket", "228": "playing cymbals", "229": "playing didgeridoo", "230": "playing drums", "231": "playing flute", "232": "playing guitar", "233": "playing harmonica", "234": "playing harp", "235": "playing ice hockey", "236": "playing keyboard", "237": "playing kickball", "238": "playing monopoly", "239": "playing organ", "240": "playing paintball", "241": "playing piano", "242": "playing poker", "243": "playing recorder", "244": "playing saxophone", "245": "playing squash or racquetball", "246": "playing tennis", "247": "playing trombone", "248": "playing trumpet", "249": "playing ukulele", "250": "playing violin", "251": "playing volleyball", "252": "playing xylophone", "253": "pole vault", "254": "presenting weather forecast", "255": "pull ups", "256": "pumping fist", "257": "pumping gas", "258": "punching bag", "259": "punching person (boxing)", "260": "push up", "261": "pushing car", "262": "pushing cart", "263": "pushing wheelchair", "264": "reading book", "265": "reading newspaper", "266": "recording music", "267": "riding a bike", "268": "riding camel", "269": "riding elephant", "270": "riding mechanical bull", "271": "riding mountain bike", "272": "riding mule", "273": "riding or walking with horse", "274": "riding scooter", "275": "riding unicycle", "276": "ripping paper", "277": "robot dancing", "278": "rock climbing", "279": "rock scissors paper", "280": "roller skating", "281": "running on treadmill", "282": "sailing", "283": "salsa dancing", "284": "sanding floor", "285": "scrambling eggs", "286": "scuba diving", "287": "setting table", "288": "shaking hands", "289": "shaking head", "290": "sharpening knives", "291": "sharpening pencil", "292": "shaving head", "293": "shaving legs", "294": "shearing sheep", "295": "shining shoes", "296": "shooting basketball", "297": "shooting goal (soccer)", "298": "shot put", "299": "shoveling snow", "300": "shredding paper", "301": "shuffling cards", "302": "side kick", "303": "sign language interpreting", "304": "singing", "305": "situp", "306": "skateboarding", "307": "ski jumping", "308": "skiing (not slalom or crosscountry)", "309": "skiing crosscountry", "310": "skiing slalom", "311": "skipping rope", "312": "skydiving", "313": "slacklining", "314": "slapping", "315": "sled dog racing", "316": "smoking", "317": "smoking hookah", "318": "snatch weight lifting", "319": "sneezing", "320": "sniffing", "321": "snorkeling", "322": "snowboarding", "323": "snowkiting", "324": "snowmobiling", "325": "somersaulting", "326": "spinning poi", "327": "spray painting", "328": "spraying", "329": "springboard diving", "330": "squat", "331": "sticking tongue out", "332": "stomping grapes", "333": "stretching arm", "334": "stretching leg", "335": "strumming guitar", "336": "surfing crowd", "337": "surfing water", "338": "sweeping floor", "339": "swimming backstroke", "340": "swimming breast stroke", "341": "swimming butterfly stroke", "342": "swing dancing", "343": "swinging legs", "344": "swinging on something", "345": "sword fighting", "346": "tai chi", "347": "taking a shower", "348": "tango dancing", "349": "tap dancing", "350": "tapping guitar", "351": "tapping pen", "352": "tasting beer", "353": "tasting food", "354": "testifying", "355": "texting", "356": "throwing axe", "357": "throwing ball", "358": "throwing discus", "359": "tickling", "360": "tobogganing", "361": "tossing coin", "362": "tossing salad", "363": "training dog", "364": "trapezing", "365": "trimming or shaving beard", "366": "trimming trees", "367": "triple jump", "368": "tying bow tie", "369": "tying knot (not on a tie)", "370": "tying tie", "371": "unboxing", "372": "unloading truck", "373": "using computer", "374": "using remote controller (not gaming)", "375": "using segway", "376": "vault", "377": "waiting in line", "378": "walking the dog", "379": "washing dishes", "380": "washing feet", "381": "washing hair", "382": "washing hands", "383": "water skiing", "384": "water sliding", "385": "watering plants", "386": "waxing back", "387": "waxing chest", "388": "waxing eyebrows", "389": "waxing legs", "390": "weaving basket", "391": "welding", "392": "whistling", "393": "windsurfing", "394": "wrapping present", "395": "wrestling", "396": "writing", "397": "yawning", "398": "yoga", "399": "zumba" }, "image_size": 224, "initializer_range": 0.02, "intermediate_size": 3072, "is_decoder": false, "is_encoder_decoder": false, "label2id": { "abseiling": 0, "air drumming": 1, "answering questions": 2, "applauding": 3, "applying cream": 4, "archery": 5, "arm wrestling": 6, "arranging flowers": 7, "assembling computer": 8, "auctioning": 9, "baby waking up": 10, "baking cookies": 11, "balloon blowing": 12, "bandaging": 13, "barbequing": 14, "bartending": 15, "beatboxing": 16, "bee keeping": 17, "belly dancing": 18, "bench pressing": 19, "bending back": 20, "bending metal": 21, "biking through snow": 22, "blasting sand": 23, "blowing glass": 24, "blowing leaves": 25, "blowing nose": 26, "blowing out candles": 27, "bobsledding": 28, "bookbinding": 29, "bouncing on trampoline": 30, "bowling": 31, "braiding hair": 32, "breading or breadcrumbing": 33, "breakdancing": 34, "brush painting": 35, "brushing hair": 36, "brushing teeth": 37, "building cabinet": 38, "building shed": 39, "bungee jumping": 40, "busking": 41, "canoeing or kayaking": 42, "capoeira": 43, "carrying baby": 44, "cartwheeling": 45, "carving pumpkin": 46, "catching fish": 47, "catching or throwing baseball": 48, "catching or throwing frisbee": 49, "catching or throwing softball": 50, "celebrating": 51, "changing oil": 52, "changing wheel": 53, "checking tires": 54, "cheerleading": 55, "chopping wood": 56, "clapping": 57, "clay pottery making": 58, "clean and jerk": 59, "cleaning floor": 60, "cleaning gutters": 61, "cleaning pool": 62, "cleaning shoes": 63, "cleaning toilet": 64, "cleaning windows": 65, "climbing a rope": 66, "climbing ladder": 67, "climbing tree": 68, "contact juggling": 69, "cooking chicken": 70, "cooking egg": 71, "cooking on campfire": 72, "cooking sausages": 73, "counting money": 74, "country line dancing": 75, "cracking neck": 76, "crawling baby": 77, "crossing river": 78, "crying": 79, "curling hair": 80, "cutting nails": 81, "cutting pineapple": 82, "cutting watermelon": 83, "dancing ballet": 84, "dancing charleston": 85, "dancing gangnam style": 86, "dancing macarena": 87, "deadlifting": 88, "decorating the christmas tree": 89, "digging": 90, "dining": 91, "disc golfing": 92, "diving cliff": 93, "dodgeball": 94, "doing aerobics": 95, "doing laundry": 96, "doing nails": 97, "drawing": 98, "dribbling basketball": 99, "drinking": 100, "drinking beer": 101, "drinking shots": 102, "driving car": 103, "driving tractor": 104, "drop kicking": 105, "drumming fingers": 106, "dunking basketball": 107, "dying hair": 108, "eating burger": 109, "eating cake": 110, "eating carrots": 111, "eating chips": 112, "eating doughnuts": 113, "eating hotdog": 114, "eating ice cream": 115, "eating spaghetti": 116, "eating watermelon": 117, "egg hunting": 118, "exercising arm": 119, "exercising with an exercise ball": 120, "extinguishing fire": 121, "faceplanting": 122, "feeding birds": 123, "feeding fish": 124, "feeding goats": 125, "filling eyebrows": 126, "finger snapping": 127, "fixing hair": 128, "flipping pancake": 129, "flying kite": 130, "folding clothes": 131, "folding napkins": 132, "folding paper": 133, "front raises": 134, "frying vegetables": 135, "garbage collecting": 136, "gargling": 137, "getting a haircut": 138, "getting a tattoo": 139, "giving or receiving award": 140, "golf chipping": 141, "golf driving": 142, "golf putting": 143, "grinding meat": 144, "grooming dog": 145, "grooming horse": 146, "gymnastics tumbling": 147, "hammer throw": 148, "headbanging": 149, "headbutting": 150, "high jump": 151, "high kick": 152, "hitting baseball": 153, "hockey stop": 154, "holding snake": 155, "hopscotch": 156, "hoverboarding": 157, "hugging": 158, "hula hooping": 159, "hurdling": 160, "hurling (sport)": 161, "ice climbing": 162, "ice fishing": 163, "ice skating": 164, "ironing": 165, "javelin throw": 166, "jetskiing": 167, "jogging": 168, "juggling balls": 169, "juggling fire": 170, "juggling soccer ball": 171, "jumping into pool": 172, "jumpstyle dancing": 173, "kicking field goal": 174, "kicking soccer ball": 175, "kissing": 176, "kitesurfing": 177, "knitting": 178, "krumping": 179, "laughing": 180, "laying bricks": 181, "long jump": 182, "lunge": 183, "making a cake": 184, "making a sandwich": 185, "making bed": 186, "making jewelry": 187, "making pizza": 188, "making snowman": 189, "making sushi": 190, "making tea": 191, "marching": 192, "massaging back": 193, "massaging feet": 194, "massaging legs": 195, "massaging person's head": 196, "milking cow": 197, "mopping floor": 198, "motorcycling": 199, "moving furniture": 200, "mowing lawn": 201, "news anchoring": 202, "opening bottle": 203, "opening present": 204, "paragliding": 205, "parasailing": 206, "parkour": 207, "passing American football (in game)": 208, "passing American football (not in game)": 209, "peeling apples": 210, "peeling potatoes": 211, "petting animal (not cat)": 212, "petting cat": 213, "picking fruit": 214, "planting trees": 215, "plastering": 216, "playing accordion": 217, "playing badminton": 218, "playing bagpipes": 219, "playing basketball": 220, "playing bass guitar": 221, "playing cards": 222, "playing cello": 223, "playing chess": 224, "playing clarinet": 225, "playing controller": 226, "playing cricket": 227, "playing cymbals": 228, "playing didgeridoo": 229, "playing drums": 230, "playing flute": 231, "playing guitar": 232, "playing harmonica": 233, "playing harp": 234, "playing ice hockey": 235, "playing keyboard": 236, "playing kickball": 237, "playing monopoly": 238, "playing organ": 239, "playing paintball": 240, "playing piano": 241, "playing poker": 242, "playing recorder": 243, "playing saxophone": 244, "playing squash or racquetball": 245, "playing tennis": 246, "playing trombone": 247, "playing trumpet": 248, "playing ukulele": 249, "playing violin": 250, "playing volleyball": 251, "playing xylophone": 252, "pole vault": 253, "presenting weather forecast": 254, "pull ups": 255, "pumping fist": 256, "pumping gas": 257, "punching bag": 258, "punching person (boxing)": 259, "push up": 260, "pushing car": 261, "pushing cart": 262, "pushing wheelchair": 263, "reading book": 264, "reading newspaper": 265, "recording music": 266, "riding a bike": 267, "riding camel": 268, "riding elephant": 269, "riding mechanical bull": 270, "riding mountain bike": 271, "riding mule": 272, "riding or walking with horse": 273, "riding scooter": 274, "riding unicycle": 275, "ripping paper": 276, "robot dancing": 277, "rock climbing": 278, "rock scissors paper": 279, "roller skating": 280, "running on treadmill": 281, "sailing": 282, "salsa dancing": 283, "sanding floor": 284, "scrambling eggs": 285, "scuba diving": 286, "setting table": 287, "shaking hands": 288, "shaking head": 289, "sharpening knives": 290, "sharpening pencil": 291, "shaving head": 292, "shaving legs": 293, "shearing sheep": 294, "shining shoes": 295, "shooting basketball": 296, "shooting goal (soccer)": 297, "shot put": 298, "shoveling snow": 299, "shredding paper": 300, "shuffling cards": 301, "side kick": 302, "sign language interpreting": 303, "singing": 304, "situp": 305, "skateboarding": 306, "ski jumping": 307, "skiing (not slalom or crosscountry)": 308, "skiing crosscountry": 309, "skiing slalom": 310, "skipping rope": 311, "skydiving": 312, "slacklining": 313, "slapping": 314, "sled dog racing": 315, "smoking": 316, "smoking hookah": 317, "snatch weight lifting": 318, "sneezing": 319, "sniffing": 320, "snorkeling": 321, "snowboarding": 322, "snowkiting": 323, "snowmobiling": 324, "somersaulting": 325, "spinning poi": 326, "spray painting": 327, "spraying": 328, "springboard diving": 329, "squat": 330, "sticking tongue out": 331, "stomping grapes": 332, "stretching arm": 333, "stretching leg": 334, "strumming guitar": 335, "surfing crowd": 336, "surfing water": 337, "sweeping floor": 338, "swimming backstroke": 339, "swimming breast stroke": 340, "swimming butterfly stroke": 341, "swing dancing": 342, "swinging legs": 343, "swinging on something": 344, "sword fighting": 345, "tai chi": 346, "taking a shower": 347, "tango dancing": 348, "tap dancing": 349, "tapping guitar": 350, "tapping pen": 351, "tasting beer": 352, "tasting food": 353, "testifying": 354, "texting": 355, "throwing axe": 356, "throwing ball": 357, "throwing discus": 358, "tickling": 359, "tobogganing": 360, "tossing coin": 361, "tossing salad": 362, "training dog": 363, "trapezing": 364, "trimming or shaving beard": 365, "trimming trees": 366, "triple jump": 367, "tying bow tie": 368, "tying knot (not on a tie)": 369, "tying tie": 370, "unboxing": 371, "unloading truck": 372, "using computer": 373, "using remote controller (not gaming)": 374, "using segway": 375, "vault": 376, "waiting in line": 377, "walking the dog": 378, "washing dishes": 379, "washing feet": 380, "washing hair": 381, "washing hands": 382, "water skiing": 383, "water sliding": 384, "watering plants": 385, "waxing back": 386, "waxing chest": 387, "waxing eyebrows": 388, "waxing legs": 389, "weaving basket": 390, "welding": 391, "whistling": 392, "windsurfing": 393, "wrapping present": 394, "wrestling": 395, "writing": 396, "yawning": 397, "yoga": 398, "zumba": 399 }, "layer_norm_eps": 1e-06, "length_penalty": 1.0, "max_length": 20, "min_length": 0, "model_type": "timesformer", "no_repeat_ngram_size": 0, "num_attention_heads": 12, "num_beam_groups": 1, "num_beams": 1, "num_channels": 3, "num_frames": 8, "num_hidden_layers": 12, "num_return_sequences": 1, "output_attentions": false, "output_hidden_states": false, "output_scores": false, "pad_token_id": null, "patch_size": 16, "prefix": null, "problem_type": null, "pruned_heads": {}, "qkv_bias": true, "remove_invalid_values": false, "repetition_penalty": 1.0, "return_dict": true, "return_dict_in_generate": false, "sep_token_id": null, "suppress_tokens": null, "task_specific_params": null, "temperature": 1.0, "tf_legacy_loss": false, "tie_encoder_decoder": false, "tie_word_embeddings": true, "tokenizer_class": null, "top_k": 50, "top_p": 1.0, "torch_dtype": "float32", "torchscript": false, "transformers_version": "4.30.2", "typical_p": 1.0, "use_bfloat16": false }, "is_encoder_decoder": true, "model_type": "vision-encoder-decoder", "pad_token_id": 0, "tie_word_embeddings": false, "torch_dtype": "float32", "transformers_version": null }