{ "_name_or_path": "UBC-NLP/serengeti", "architectures": [ "XLMRobertaForSequenceClassification" ], "attention_probs_dropout_prob": 0.1, "bos_token_id": 0, "classifier_dropout": null, "eos_token_id": 2, "finetuning_task": "text-classification", "hidden_act": "gelu", "hidden_dropout_prob": 0.1, "hidden_size": 768, "id2label": { "0": "aar", "1": "aba", "2": "abn", "3": "acd", "4": "ach", "5": "ada", "6": "adh", "7": "adj", "8": "afr", "9": "agq", "10": "aha", "11": "ajg", "12": "akp", "13": "alz", "14": "amh", "15": "ann", "16": "anu", "17": "anv", "18": "asa", "19": "asg", "20": "atg", "21": "ati", "22": "avn", "23": "avu", "24": "azo", "25": "bam", "26": "bav", "27": "bba", "28": "bbj", "29": "bbk", "30": "bci", "31": "bcn", "32": "bcw", "33": "bcy", "34": "bdh", "35": "bds", "36": "bem", "37": "beq", "38": "ber", "39": "bex", "40": "bez", "41": "bfa", "42": "bfd", "43": "bfo", "44": "bib", "45": "bim", "46": "bin", "47": "biv", "48": "bjv", "49": "bky", "50": "bmo", "51": "bmv", "52": "bom", "53": "bov", "54": "box", "55": "bqc", "56": "bqj", "57": "bsc", "58": "bsp", "59": "bss", "60": "bst", "61": "bud", "62": "bum", "63": "bun", "64": "bus", "65": "buy", "66": "bwr", "67": "bwu", "68": "bxk", "69": "byf", "70": "byv", "71": "bza", "72": "bzw", "73": "cce", "74": "chw", "75": "cjk", "76": "cko", "77": "cme", "78": "cop", "79": "cou", "80": "crs", "81": "csk", "82": "cwe", "83": "daa", "84": "dag", "85": "dav", "86": "dga", "87": "dgd", "88": "dgi", "89": "dhm", "90": "dib", "91": "did", "92": "dig", "93": "dik", "94": "dip", "95": "diu", "96": "dks", "97": "dnj", "98": "dow", "99": "dsh", "100": "dua", "101": "dug", "102": "dwr", "103": "dyi", "104": "dyu", "105": "ebr", "106": "ebu", "107": "efi", "108": "ego", "109": "eka", "110": "eko", "111": "eto", "112": "etu", "113": "etx", "114": "ewe", "115": "ewo", "116": "fak", "117": "fat", "118": "ffm", "119": "fia", "120": "fip", "121": "flr", "122": "fon", "123": "fub", "124": "fue", "125": "fuf", "126": "fuh", "127": "ful", "128": "fuq", "129": "fuv", "130": "gaa", "131": "gax", "132": "gaz", "133": "gbo", "134": "gbr", "135": "gde", "136": "gid", "137": "giz", "138": "gjn", "139": "gkn", "140": "gkp", "141": "gmv", "142": "gna", "143": "gnd", "144": "gng", "145": "gof", "146": "gog", "147": "gol", "148": "gqr", "149": "gso", "150": "gud", "151": "gur", "152": "guw", "153": "gux", "154": "guz", "155": "gvl", "156": "gwr", "157": "gya", "158": "hag", "159": "har", "160": "hau", "161": "hay", "162": "hbb", "163": "heh", "164": "her", "165": "hgm", "166": "hna", "167": "ibb", "168": "ibo", "169": "idu", "170": "igb", "171": "ige", "172": "igl", "173": "ijn", "174": "ikk", "175": "ikw", "176": "iqw", "177": "iri", "178": "ish", "179": "iso", "180": "iyx", "181": "izr", "182": "izz", "183": "jgo", "184": "jib", "185": "jit", "186": "jmc", "187": "kab", "188": "kam", "189": "kbn", "190": "kbo", "191": "kbp", "192": "kby", "193": "kcg", "194": "kck", "195": "kdc", "196": "kde", "197": "kdh", "198": "kdi", "199": "kdj", "200": "kdl", "201": "kdn", "202": "kea", "203": "ken", "204": "khy", "205": "kia", "206": "kik", "207": "kin", "208": "kiz", "209": "kki", "210": "kkj", "211": "kln", "212": "klu", "213": "kma", "214": "kmb", "215": "kmy", "216": "knf", "217": "kng", "218": "knk", "219": "kno", "220": "koo", "221": "koq", "222": "kqn", "223": "kqp", "224": "kqs", "225": "kqy", "226": "kri", "227": "krs", "228": "krw", "229": "krx", "230": "ksb", "231": "ksf", "232": "ksp", "233": "ktj", "234": "ktu", "235": "kua", "236": "kub", "237": "kuj", "238": "kus", "239": "kvj", "240": "kwn", "241": "kyf", "242": "kyq", "243": "kzr", "244": "lai", "245": "laj", "246": "lam", "247": "lap", "248": "lee", "249": "lef", "250": "lem", "251": "lgg", "252": "lgm", "253": "lia", "254": "lik", "255": "lin", "256": "lip", "257": "lmd", "258": "lmp", "259": "lnl", "260": "log", "261": "lom", "262": "loq", "263": "lot", "264": "loz", "265": "lro", "266": "lsm", "267": "lth", "268": "lto", "269": "lua", "270": "luc", "271": "lue", "272": "lug", "273": "lun", "274": "luo", "275": "lwg", "276": "lwo", "277": "maf", "278": "mas", "279": "maw", "280": "mbu", "281": "mck", "282": "mcn", "283": "mcp", "284": "mcu", "285": "mda", "286": "mdm", "287": "mdy", "288": "men", "289": "meq", "290": "mer", "291": "mev", "292": "mfe", "293": "mfg", "294": "mfh", "295": "mfi", "296": "mfk", "297": "mfq", "298": "mfz", "299": "mgc", "300": "mgh", "301": "mgo", "302": "mgq", "303": "mgr", "304": "mgw", "305": "mif", "306": "mkl", "307": "mlg", "308": "mlr", "309": "mmy", "310": "mnf", "311": "mnk", "312": "moa", "313": "mos", "314": "moy", "315": "moz", "316": "mpe", "317": "mpg", "318": "mqb", "319": "msc", "320": "mur", "321": "muy", "322": "mwe", "323": "mwm", "324": "mwn", "325": "mws", "326": "myb", "327": "myk", "328": "myx", "329": "mzm", "330": "mzw", "331": "nan_lang", "332": "naq", "333": "naw", "334": "nba", "335": "nbl", "336": "ncu", "337": "ndc", "338": "nde", "339": "ndh", "340": "ndj", "341": "ndo", "342": "ndv", "343": "ndz", "344": "ngb", "345": "ngc", "346": "ngl", "347": "ngn", "348": "ngo", "349": "ngp", "350": "nhr", "351": "nhu", "352": "nih", "353": "nim", "354": "nin", "355": "niy", "356": "nka", "357": "nko", "358": "nla", "359": "nnb", "360": "nnh", "361": "nnq", "362": "nnw", "363": "nse", "364": "nso", "365": "ntr", "366": "nuj", "367": "nus", "368": "nwb", "369": "nxd", "370": "nya", "371": "nyb", "372": "nyd", "373": "nyf", "374": "nyk", "375": "nym", "376": "nyn", "377": "nyo", "378": "nyu", "379": "nyy", "380": "nza", "381": "nzi", "382": "odu", "383": "ogo", "384": "oke", "385": "okr", "386": "oku", "387": "orm", "388": "ozm", "389": "pcm", "390": "pem", "391": "pkb", "392": "pko", "393": "pov", "394": "poy", "395": "rag", "396": "rel", "397": "rif", "398": "rim", "399": "rnd", "400": "rng", "401": "rub", "402": "run", "403": "rwk", "404": "sag", "405": "saq", "406": "sba", "407": "sbd", "408": "sbp", "409": "sbs", "410": "sby", "411": "sef", "412": "ses", "413": "sev", "414": "sfw", "415": "sgw", "416": "shi", "417": "shj", "418": "shk", "419": "sid", "420": "sig", "421": "sil", "422": "sna", "423": "snf", "424": "sng", "425": "snw", "426": "som", "427": "sop", "428": "sor", "429": "sot", "430": "soy", "431": "spp", "432": "ssw", "433": "suk", "434": "sus", "435": "swa", "436": "swc", "437": "swh", "438": "swk", "439": "sxb", "440": "taq", "441": "tcc", "442": "tcd", "443": "ted", "444": "tem", "445": "teo", "446": "tex", "447": "tgw", "448": "thk", "449": "thv", "450": "tir", "451": "tiv", "452": "tke", "453": "tlj", "454": "tll", "455": "tog", "456": "toh", "457": "toi", "458": "tpm", "459": "tsc", "460": "tsn", "461": "tso", "462": "tsw", "463": "ttj", "464": "ttq", "465": "ttr", "466": "tui", "467": "tul", "468": "tum", "469": "tuv", "470": "tvu", "471": "twi", "472": "umb", "473": "urh", "474": "uth", "475": "vag", "476": "vai", "477": "ven", "478": "vid", "479": "vif", "480": "vmk", "481": "vmw", "482": "vun", "483": "vut", "484": "wal", "485": "wbi", "486": "wec", "487": "wes", "488": "wib", "489": "wmw", "490": "wol", "491": "won", "492": "xan", "493": "xed", "494": "xho", "495": "xnz", "496": "xog", "497": "xon", "498": "xpe", "499": "xrb", "500": "xsm", "501": "xtc", "502": "xuo", "503": "yal", "504": "yam", "505": "yao", "506": "yat", "507": "yba", "508": "ybb", "509": "yom", "510": "yor", "511": "yre", "512": "zaj", "513": "zdj", "514": "zga", "515": "ziw", "516": "zne", "517": "zul" }, "initializer_range": 0.02, "intermediate_size": 3072, "label2id": { "aar": 0, "aba": 1, "abn": 2, "acd": 3, "ach": 4, "ada": 5, "adh": 6, "adj": 7, "afr": 8, "agq": 9, "aha": 10, "ajg": 11, "akp": 12, "alz": 13, "amh": 14, "ann": 15, "anu": 16, "anv": 17, "asa": 18, "asg": 19, "atg": 20, "ati": 21, "avn": 22, "avu": 23, "azo": 24, "bam": 25, "bav": 26, "bba": 27, "bbj": 28, "bbk": 29, "bci": 30, "bcn": 31, "bcw": 32, "bcy": 33, "bdh": 34, "bds": 35, "bem": 36, "beq": 37, "ber": 38, "bex": 39, "bez": 40, "bfa": 41, "bfd": 42, "bfo": 43, "bib": 44, "bim": 45, "bin": 46, "biv": 47, "bjv": 48, "bky": 49, "bmo": 50, "bmv": 51, "bom": 52, "bov": 53, "box": 54, "bqc": 55, "bqj": 56, "bsc": 57, "bsp": 58, "bss": 59, "bst": 60, "bud": 61, "bum": 62, "bun": 63, "bus": 64, "buy": 65, "bwr": 66, "bwu": 67, "bxk": 68, "byf": 69, "byv": 70, "bza": 71, "bzw": 72, "cce": 73, "chw": 74, "cjk": 75, "cko": 76, "cme": 77, "cop": 78, "cou": 79, "crs": 80, "csk": 81, "cwe": 82, "daa": 83, "dag": 84, "dav": 85, "dga": 86, "dgd": 87, "dgi": 88, "dhm": 89, "dib": 90, "did": 91, "dig": 92, "dik": 93, "dip": 94, "diu": 95, "dks": 96, "dnj": 97, "dow": 98, "dsh": 99, "dua": 100, "dug": 101, "dwr": 102, "dyi": 103, "dyu": 104, "ebr": 105, "ebu": 106, "efi": 107, "ego": 108, "eka": 109, "eko": 110, "eto": 111, "etu": 112, "etx": 113, "ewe": 114, "ewo": 115, "fak": 116, "fat": 117, "ffm": 118, "fia": 119, "fip": 120, "flr": 121, "fon": 122, "fub": 123, "fue": 124, "fuf": 125, "fuh": 126, "ful": 127, "fuq": 128, "fuv": 129, "gaa": 130, "gax": 131, "gaz": 132, "gbo": 133, "gbr": 134, "gde": 135, "gid": 136, "giz": 137, "gjn": 138, "gkn": 139, "gkp": 140, "gmv": 141, "gna": 142, "gnd": 143, "gng": 144, "gof": 145, "gog": 146, "gol": 147, "gqr": 148, "gso": 149, "gud": 150, "gur": 151, "guw": 152, "gux": 153, "guz": 154, "gvl": 155, "gwr": 156, "gya": 157, "hag": 158, "har": 159, "hau": 160, "hay": 161, "hbb": 162, "heh": 163, "her": 164, "hgm": 165, "hna": 166, "ibb": 167, "ibo": 168, "idu": 169, "igb": 170, "ige": 171, "igl": 172, "ijn": 173, "ikk": 174, "ikw": 175, "iqw": 176, "iri": 177, "ish": 178, "iso": 179, "iyx": 180, "izr": 181, "izz": 182, "jgo": 183, "jib": 184, "jit": 185, "jmc": 186, "kab": 187, "kam": 188, "kbn": 189, "kbo": 190, "kbp": 191, "kby": 192, "kcg": 193, "kck": 194, "kdc": 195, "kde": 196, "kdh": 197, "kdi": 198, "kdj": 199, "kdl": 200, "kdn": 201, "kea": 202, "ken": 203, "khy": 204, "kia": 205, "kik": 206, "kin": 207, "kiz": 208, "kki": 209, "kkj": 210, "kln": 211, "klu": 212, "kma": 213, "kmb": 214, "kmy": 215, "knf": 216, "kng": 217, "knk": 218, "kno": 219, "koo": 220, "koq": 221, "kqn": 222, "kqp": 223, "kqs": 224, "kqy": 225, "kri": 226, "krs": 227, "krw": 228, "krx": 229, "ksb": 230, "ksf": 231, "ksp": 232, "ktj": 233, "ktu": 234, "kua": 235, "kub": 236, "kuj": 237, "kus": 238, "kvj": 239, "kwn": 240, "kyf": 241, "kyq": 242, "kzr": 243, "lai": 244, "laj": 245, "lam": 246, "lap": 247, "lee": 248, "lef": 249, "lem": 250, "lgg": 251, "lgm": 252, "lia": 253, "lik": 254, "lin": 255, "lip": 256, "lmd": 257, "lmp": 258, "lnl": 259, "log": 260, "lom": 261, "loq": 262, "lot": 263, "loz": 264, "lro": 265, "lsm": 266, "lth": 267, "lto": 268, "lua": 269, "luc": 270, "lue": 271, "lug": 272, "lun": 273, "luo": 274, "lwg": 275, "lwo": 276, "maf": 277, "mas": 278, "maw": 279, "mbu": 280, "mck": 281, "mcn": 282, "mcp": 283, "mcu": 284, "mda": 285, "mdm": 286, "mdy": 287, "men": 288, "meq": 289, "mer": 290, "mev": 291, "mfe": 292, "mfg": 293, "mfh": 294, "mfi": 295, "mfk": 296, "mfq": 297, "mfz": 298, "mgc": 299, "mgh": 300, "mgo": 301, "mgq": 302, "mgr": 303, "mgw": 304, "mif": 305, "mkl": 306, "mlg": 307, "mlr": 308, "mmy": 309, "mnf": 310, "mnk": 311, "moa": 312, "mos": 313, "moy": 314, "moz": 315, "mpe": 316, "mpg": 317, "mqb": 318, "msc": 319, "mur": 320, "muy": 321, "mwe": 322, "mwm": 323, "mwn": 324, "mws": 325, "myb": 326, "myk": 327, "myx": 328, "mzm": 329, "mzw": 330, "nan_lang": 331, "naq": 332, "naw": 333, "nba": 334, "nbl": 335, "ncu": 336, "ndc": 337, "nde": 338, "ndh": 339, "ndj": 340, "ndo": 341, "ndv": 342, "ndz": 343, "ngb": 344, "ngc": 345, "ngl": 346, "ngn": 347, "ngo": 348, "ngp": 349, "nhr": 350, "nhu": 351, "nih": 352, "nim": 353, "nin": 354, "niy": 355, "nka": 356, "nko": 357, "nla": 358, "nnb": 359, "nnh": 360, "nnq": 361, "nnw": 362, "nse": 363, "nso": 364, "ntr": 365, "nuj": 366, "nus": 367, "nwb": 368, "nxd": 369, "nya": 370, "nyb": 371, "nyd": 372, "nyf": 373, "nyk": 374, "nym": 375, "nyn": 376, "nyo": 377, "nyu": 378, "nyy": 379, "nza": 380, "nzi": 381, "odu": 382, "ogo": 383, "oke": 384, "okr": 385, "oku": 386, "orm": 387, "ozm": 388, "pcm": 389, "pem": 390, "pkb": 391, "pko": 392, "pov": 393, "poy": 394, "rag": 395, "rel": 396, "rif": 397, "rim": 398, "rnd": 399, "rng": 400, "rub": 401, "run": 402, "rwk": 403, "sag": 404, "saq": 405, "sba": 406, "sbd": 407, "sbp": 408, "sbs": 409, "sby": 410, "sef": 411, "ses": 412, "sev": 413, "sfw": 414, "sgw": 415, "shi": 416, "shj": 417, "shk": 418, "sid": 419, "sig": 420, "sil": 421, "sna": 422, "snf": 423, "sng": 424, "snw": 425, "som": 426, "sop": 427, "sor": 428, "sot": 429, "soy": 430, "spp": 431, "ssw": 432, "suk": 433, "sus": 434, "swa": 435, "swc": 436, "swh": 437, "swk": 438, "sxb": 439, "taq": 440, "tcc": 441, "tcd": 442, "ted": 443, "tem": 444, "teo": 445, "tex": 446, "tgw": 447, "thk": 448, "thv": 449, "tir": 450, "tiv": 451, "tke": 452, "tlj": 453, "tll": 454, "tog": 455, "toh": 456, "toi": 457, "tpm": 458, "tsc": 459, "tsn": 460, "tso": 461, "tsw": 462, "ttj": 463, "ttq": 464, "ttr": 465, "tui": 466, "tul": 467, "tum": 468, "tuv": 469, "tvu": 470, "twi": 471, "umb": 472, "urh": 473, "uth": 474, "vag": 475, "vai": 476, "ven": 477, "vid": 478, "vif": 479, "vmk": 480, "vmw": 481, "vun": 482, "vut": 483, "wal": 484, "wbi": 485, "wec": 486, "wes": 487, "wib": 488, "wmw": 489, "wol": 490, "won": 491, "xan": 492, "xed": 493, "xho": 494, "xnz": 495, "xog": 496, "xon": 497, "xpe": 498, "xrb": 499, "xsm": 500, "xtc": 501, "xuo": 502, "yal": 503, "yam": 504, "yao": 505, "yat": 506, "yba": 507, "ybb": 508, "yom": 509, "yor": 510, "yre": 511, "zaj": 512, "zdj": 513, "zga": 514, "ziw": 515, "zne": 516, "zul": 517 }, "layer_norm_eps": 1e-05, "max_position_embeddings": 514, "model_type": "xlm-roberta", "num_attention_heads": 12, "num_hidden_layers": 12, "output_past": true, "pad_token_id": 1, "position_embedding_type": "absolute", "problem_type": "single_label_classification", "torch_dtype": "float32", "transformers_version": "4.31.0", "type_vocab_size": 1, "use_cache": true, "vocab_size": 250004 }