{ "_name_or_path": "model/xlm-roberta-base-finetuned-language-detection-tatoeba/checkpoint-32346/", "architectures": [ "XLMRobertaForSequenceClassification" ], "attention_probs_dropout_prob": 0.1, "bos_token_id": 0, "classifier_dropout": null, "eos_token_id": 2, "hidden_act": "gelu", "hidden_dropout_prob": 0.1, "hidden_size": 768, "id2label": { "0": "Ainu", "1": "Algerian Arabic", "2": "Asturian", "3": "Awadhi", "4": "Baluchi", "5": "Banjar", "6": "Bavarian", "7": "Berber languages", "8": "Brithenig", "9": "Buriat", "10": "Cantonese", "11": "Cebuano", "12": "Central Dusun", "13": "Central Kanuri", "14": "Choctaw", "15": "Chukot", "16": "Coastal Kadazan", "17": "Congo Swahili", "18": "Creek", "19": "Crimean Turkish", "20": "Eastern Mari", "21": "Egyptian Arabic", "22": "Emilian-Romagnol", "23": "Evenki", "24": "Gondi", "25": "Guadeloupean Creole French", "26": "Gulf Arabic", "27": "Hawaiian", "28": "Ho", "29": "Hunsrik", "30": "Iloko", "31": "Iranian Persian", "32": "Kabyle", "33": "Kalmyk", "34": "Karachay-Balkar", "35": "Karelian", "36": "Kashubian", "37": "Khasi", "38": "Khmer", "39": "Kumyk", "40": "Kven Finnish", "41": "Kyrgyz", "42": "Ladino", "43": "Latgalian", "44": "Lhunda", "45": "Ligurian", "46": "Lingua Franca Nova", "47": "Low German", "48": "Lower Sorbian", "49": "Malay (individual language)", "50": "Mikasuki", "51": "Minangkabau", "52": "Mingrelian", "53": "Modern Greek", "54": "Nahuatl languages", "55": "Nepali (individual language)", "56": "Nogai", "57": "North Frisian", "58": "North Levantine Arabic", "59": "North Moluccan Malay", "60": "Northern Kurdish", "61": "Nuer", "62": "Occitan", "63": "Old Tupi", "64": "Oriya", "65": "Pampanga", "66": "Pattani Malay", "67": "Picard", "68": "Piedmontese", "69": "Punjabi", "70": "Rohingya", "71": "Romany", "72": "Rusyn", "73": "Sakha", "74": "Santali", "75": "Seraiki", "76": "Sorani Kurdish", "77": "Southern Kurdish", "78": "Sranan Tongo", "79": "Standard Latvian", "80": "Standard Malay", "81": "Standard Moroccan Tamazight", "82": "Swabian", "83": "Swahili", "84": "Swiss German", "85": "Tachawit", "86": "Tachelhit", "87": "Tarifit", "88": "Tase Naga", "89": "Tetum", "90": "Tigr\u00e9", "91": "Tok Pisin", "92": "Udmurt", "93": "Upper Sorbian", "94": "Uyghur", "95": "Venetian", "96": "Waray-Waray", "97": "Wayuu", "98": "Wu Chinese", "99": "Zamboanga Chavacano", "100": "Zaza", "101": "af", "102": "am", "103": "an", "104": "ar", "105": "as", "106": "az", "107": "ba", "108": "be", "109": "bg", "110": "bn", "111": "br", "112": "bs", "113": "ca", "114": "ch", "115": "cs", "116": "cv", "117": "cy", "118": "da", "119": "de", "120": "el", "121": "en", "122": "es", "123": "et", "124": "eu", "125": "fi", "126": "fo", "127": "fr", "128": "fy", "129": "ga", "130": "gd", "131": "gl", "132": "gn", "133": "gu", "134": "ha", "135": "he", "136": "hi", "137": "hr", "138": "hu", "139": "hy", "140": "id", "141": "is", "142": "it", "143": "ja", "144": "jv", "145": "ka", "146": "kk", "147": "kl", "148": "kn", "149": "ko", "150": "kw", "151": "lb", "152": "ln", "153": "lo", "154": "lt", "155": "mi", "156": "mk", "157": "ml", "158": "mn", "159": "mr", "160": "mt", "161": "my", "162": "nb", "163": "nl", "164": "nn", "165": "os", "166": "pl", "167": "pt", "168": "qu", "169": "rn", "170": "ro", "171": "ru", "172": "se", "173": "sk", "174": "sl", "175": "sq", "176": "sr", "177": "sv", "178": "sw", "179": "ta", "180": "te", "181": "th", "182": "ti", "183": "tk", "184": "tl", "185": "tr", "186": "tt", "187": "ty", "188": "uk", "189": "ur", "190": "uz", "191": "vi", "192": "wo", "193": "xh", "194": "yi", "195": "zh" }, "initializer_range": 0.02, "intermediate_size": 3072, "label2id": { "Ainu": 0, "Algerian Arabic": 1, "Asturian": 2, "Awadhi": 3, "Baluchi": 4, "Banjar": 5, "Bavarian": 6, "Berber languages": 7, "Brithenig": 8, "Buriat": 9, "Cantonese": 10, "Cebuano": 11, "Central Dusun": 12, "Central Kanuri": 13, "Choctaw": 14, "Chukot": 15, "Coastal Kadazan": 16, "Congo Swahili": 17, "Creek": 18, "Crimean Turkish": 19, "Eastern Mari": 20, "Egyptian Arabic": 21, "Emilian-Romagnol": 22, "Evenki": 23, "Gondi": 24, "Guadeloupean Creole French": 25, "Gulf Arabic": 26, "Hawaiian": 27, "Ho": 28, "Hunsrik": 29, "Iloko": 30, "Iranian Persian": 31, "Kabyle": 32, "Kalmyk": 33, "Karachay-Balkar": 34, "Karelian": 35, "Kashubian": 36, "Khasi": 37, "Khmer": 38, "Kumyk": 39, "Kven Finnish": 40, "Kyrgyz": 41, "Ladino": 42, "Latgalian": 43, "Lhunda": 44, "Ligurian": 45, "Lingua Franca Nova": 46, "Low German": 47, "Lower Sorbian": 48, "Malay (individual language)": 49, "Mikasuki": 50, "Minangkabau": 51, "Mingrelian": 52, "Modern Greek": 53, "Nahuatl languages": 54, "Nepali (individual language)": 55, "Nogai": 56, "North Frisian": 57, "North Levantine Arabic": 58, "North Moluccan Malay": 59, "Northern Kurdish": 60, "Nuer": 61, "Occitan": 62, "Old Tupi": 63, "Oriya": 64, "Pampanga": 65, "Pattani Malay": 66, "Picard": 67, "Piedmontese": 68, "Punjabi": 69, "Rohingya": 70, "Romany": 71, "Rusyn": 72, "Sakha": 73, "Santali": 74, "Seraiki": 75, "Sorani Kurdish": 76, "Southern Kurdish": 77, "Sranan Tongo": 78, "Standard Latvian": 79, "Standard Malay": 80, "Standard Moroccan Tamazight": 81, "Swabian": 82, "Swahili": 83, "Swiss German": 84, "Tachawit": 85, "Tachelhit": 86, "Tarifit": 87, "Tase Naga": 88, "Tetum": 89, "Tigr\u00e9": 90, "Tok Pisin": 91, "Udmurt": 92, "Upper Sorbian": 93, "Uyghur": 94, "Venetian": 95, "Waray-Waray": 96, "Wayuu": 97, "Wu Chinese": 98, "Zamboanga Chavacano": 99, "Zaza": 100, "af": 101, "am": 102, "an": 103, "ar": 104, "as": 105, "az": 106, "ba": 107, "be": 108, "bg": 109, "bn": 110, "br": 111, "bs": 112, "ca": 113, "ch": 114, "cs": 115, "cv": 116, "cy": 117, "da": 118, "de": 119, "el": 120, "en": 121, "es": 122, "et": 123, "eu": 124, "fi": 125, "fo": 126, "fr": 127, "fy": 128, "ga": 129, "gd": 130, "gl": 131, "gn": 132, "gu": 133, "ha": 134, "he": 135, "hi": 136, "hr": 137, "hu": 138, "hy": 139, "id": 140, "is": 141, "it": 142, "ja": 143, "jv": 144, "ka": 145, "kk": 146, "kl": 147, "kn": 148, "ko": 149, "kw": 150, "lb": 151, "ln": 152, "lo": 153, "lt": 154, "mi": 155, "mk": 156, "ml": 157, "mn": 158, "mr": 159, "mt": 160, "my": 161, "nb": 162, "nl": 163, "nn": 164, "os": 165, "pl": 166, "pt": 167, "qu": 168, "rn": 169, "ro": 170, "ru": 171, "se": 172, "sk": 173, "sl": 174, "sq": 175, "sr": 176, "sv": 177, "sw": 178, "ta": 179, "te": 180, "th": 181, "ti": 182, "tk": 183, "tl": 184, "tr": 185, "tt": 186, "ty": 187, "uk": 188, "ur": 189, "uz": 190, "vi": 191, "wo": 192, "xh": 193, "yi": 194, "zh": 195 }, "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": "4.27.3", "type_vocab_size": 1, "use_cache": true, "vocab_size": 250002 }