Upload XLMRobertaForSequenceClassification

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "model/xlm-roberta-base-finetuned-language-detection-tatoeba/checkpoint-32346/",
+  "_name_or_path": "xlm-roberta-base",
   "architectures": [
     "XLMRobertaForSequenceClassification"
   ],
@@ -11,402 +11,402 @@
   "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"
+    "0": "Ho",
+    "1": "af",
+    "2": "afb",
+    "3": "ain",
+    "4": "am",
+    "5": "an",
+    "6": "apc",
+    "7": "ar",
+    "8": "arq",
+    "9": "arz",
+    "10": "as",
+    "11": "ast",
+    "12": "awa",
+    "13": "az",
+    "14": "ba",
+    "15": "bal",
+    "16": "bar",
+    "17": "be",
+    "18": "ber",
+    "19": "bg",
+    "20": "bjn",
+    "21": "bn",
+    "22": "br",
+    "23": "bs",
+    "24": "bua",
+    "25": "bzt",
+    "26": "ca",
+    "27": "cbk",
+    "28": "ceb",
+    "29": "ch",
+    "30": "cho",
+    "31": "ckt",
+    "32": "crh",
+    "33": "cs",
+    "34": "csb",
+    "35": "cv",
+    "36": "cy",
+    "37": "da",
+    "38": "de",
+    "39": "dsb",
+    "40": "dtp",
+    "41": "egl",
+    "42": "ell",
+    "43": "en",
+    "44": "es",
+    "45": "et",
+    "46": "eu",
+    "47": "evn",
+    "48": "fas",
+    "49": "fi",
+    "50": "fkv",
+    "51": "fo",
+    "52": "fr",
+    "53": "frr",
+    "54": "fy",
+    "55": "ga",
+    "56": "gcf",
+    "57": "gd",
+    "58": "gl",
+    "59": "gn",
+    "60": "gos",
+    "61": "gsw",
+    "62": "gu",
+    "63": "guc",
+    "64": "ha",
+    "65": "haw",
+    "66": "he",
+    "67": "hi",
+    "68": "hr",
+    "69": "hrx",
+    "70": "hsb",
+    "71": "hu",
+    "72": "hy",
+    "73": "id",
+    "74": "ilo",
+    "75": "is",
+    "76": "it",
+    "77": "ja",
+    "78": "jv",
+    "79": "ka",
+    "80": "kab",
+    "81": "kha",
+    "82": "kk",
+    "83": "kl",
+    "84": "km",
+    "85": "kmr",
+    "86": "kn",
+    "87": "knc",
+    "88": "ko",
+    "89": "krc",
+    "90": "krl",
+    "91": "ku",
+    "92": "kum",
+    "93": "kw",
+    "94": "ky",
+    "95": "kzj",
+    "96": "lad",
+    "97": "lat",
+    "98": "lb",
+    "99": "lfn",
+    "100": "lij",
+    "101": "ln",
+    "102": "lo",
+    "103": "lt",
+    "104": "ltg",
+    "105": "lv",
+    "106": "max",
+    "107": "mfa",
+    "108": "mhr",
+    "109": "mi",
+    "110": "mik",
+    "111": "min",
+    "112": "mk",
+    "113": "ml",
+    "114": "mn",
+    "115": "mr",
+    "116": "ms",
+    "117": "mt",
+    "118": "mus",
+    "119": "my",
+    "120": "nah",
+    "121": "nds",
+    "122": "ne",
+    "123": "nl",
+    "124": "nn",
+    "125": "nnb",
+    "126": "nog",
+    "127": "nst",
+    "128": "nus",
+    "129": "oc",
+    "130": "or",
+    "131": "os",
+    "132": "pa",
+    "133": "pam",
+    "134": "pcd",
+    "135": "pes",
+    "136": "pl",
+    "137": "pms",
+    "138": "pt",
+    "139": "qu",
+    "140": "rhg",
+    "141": "rif",
+    "142": "rn",
+    "143": "ro",
+    "144": "rom",
+    "145": "ru",
+    "146": "rue",
+    "147": "sah",
+    "148": "sat",
+    "149": "sco",
+    "150": "sdh",
+    "151": "se",
+    "152": "shi",
+    "153": "shy",
+    "154": "sk",
+    "155": "skr",
+    "156": "sl",
+    "157": "sq",
+    "158": "sr",
+    "159": "srn",
+    "160": "sv",
+    "161": "swc",
+    "162": "swg",
+    "163": "swh",
+    "164": "ta",
+    "165": "te",
+    "166": "tet",
+    "167": "th",
+    "168": "thv",
+    "169": "ti",
+    "170": "tig",
+    "171": "tk",
+    "172": "tl",
+    "173": "tok",
+    "174": "tpw",
+    "175": "tr",
+    "176": "tt",
+    "177": "udm",
+    "178": "uig",
+    "179": "uk",
+    "180": "ur",
+    "181": "uz",
+    "182": "vec",
+    "183": "vi",
+    "184": "war",
+    "185": "wo",
+    "186": "wuu",
+    "187": "xal",
+    "188": "xh",
+    "189": "xmf",
+    "190": "yi",
+    "191": "yue",
+    "192": "zgh",
+    "193": "zh-Hant",
+    "194": "zlm",
+    "195": "zza"
   },
   "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
+    "Ho": 0,
+    "af": 1,
+    "afb": 2,
+    "ain": 3,
+    "am": 4,
+    "an": 5,
+    "apc": 6,
+    "ar": 7,
+    "arq": 8,
+    "arz": 9,
+    "as": 10,
+    "ast": 11,
+    "awa": 12,
+    "az": 13,
+    "ba": 14,
+    "bal": 15,
+    "bar": 16,
+    "be": 17,
+    "ber": 18,
+    "bg": 19,
+    "bjn": 20,
+    "bn": 21,
+    "br": 22,
+    "bs": 23,
+    "bua": 24,
+    "bzt": 25,
+    "ca": 26,
+    "cbk": 27,
+    "ceb": 28,
+    "ch": 29,
+    "cho": 30,
+    "ckt": 31,
+    "crh": 32,
+    "cs": 33,
+    "csb": 34,
+    "cv": 35,
+    "cy": 36,
+    "da": 37,
+    "de": 38,
+    "dsb": 39,
+    "dtp": 40,
+    "egl": 41,
+    "ell": 42,
+    "en": 43,
+    "es": 44,
+    "et": 45,
+    "eu": 46,
+    "evn": 47,
+    "fas": 48,
+    "fi": 49,
+    "fkv": 50,
+    "fo": 51,
+    "fr": 52,
+    "frr": 53,
+    "fy": 54,
+    "ga": 55,
+    "gcf": 56,
+    "gd": 57,
+    "gl": 58,
+    "gn": 59,
+    "gos": 60,
+    "gsw": 61,
+    "gu": 62,
+    "guc": 63,
+    "ha": 64,
+    "haw": 65,
+    "he": 66,
+    "hi": 67,
+    "hr": 68,
+    "hrx": 69,
+    "hsb": 70,
+    "hu": 71,
+    "hy": 72,
+    "id": 73,
+    "ilo": 74,
+    "is": 75,
+    "it": 76,
+    "ja": 77,
+    "jv": 78,
+    "ka": 79,
+    "kab": 80,
+    "kha": 81,
+    "kk": 82,
+    "kl": 83,
+    "km": 84,
+    "kmr": 85,
+    "kn": 86,
+    "knc": 87,
+    "ko": 88,
+    "krc": 89,
+    "krl": 90,
+    "ku": 91,
+    "kum": 92,
+    "kw": 93,
+    "ky": 94,
+    "kzj": 95,
+    "lad": 96,
+    "lat": 97,
+    "lb": 98,
+    "lfn": 99,
+    "lij": 100,
+    "ln": 101,
+    "lo": 102,
+    "lt": 103,
+    "ltg": 104,
+    "lv": 105,
+    "max": 106,
+    "mfa": 107,
+    "mhr": 108,
+    "mi": 109,
+    "mik": 110,
+    "min": 111,
+    "mk": 112,
+    "ml": 113,
+    "mn": 114,
+    "mr": 115,
+    "ms": 116,
+    "mt": 117,
+    "mus": 118,
+    "my": 119,
+    "nah": 120,
+    "nds": 121,
+    "ne": 122,
+    "nl": 123,
+    "nn": 124,
+    "nnb": 125,
+    "nog": 126,
+    "nst": 127,
+    "nus": 128,
+    "oc": 129,
+    "or": 130,
+    "os": 131,
+    "pa": 132,
+    "pam": 133,
+    "pcd": 134,
+    "pes": 135,
+    "pl": 136,
+    "pms": 137,
+    "pt": 138,
+    "qu": 139,
+    "rhg": 140,
+    "rif": 141,
+    "rn": 142,
+    "ro": 143,
+    "rom": 144,
+    "ru": 145,
+    "rue": 146,
+    "sah": 147,
+    "sat": 148,
+    "sco": 149,
+    "sdh": 150,
+    "se": 151,
+    "shi": 152,
+    "shy": 153,
+    "sk": 154,
+    "skr": 155,
+    "sl": 156,
+    "sq": 157,
+    "sr": 158,
+    "srn": 159,
+    "sv": 160,
+    "swc": 161,
+    "swg": 162,
+    "swh": 163,
+    "ta": 164,
+    "te": 165,
+    "tet": 166,
+    "th": 167,
+    "thv": 168,
+    "ti": 169,
+    "tig": 170,
+    "tk": 171,
+    "tl": 172,
+    "tok": 173,
+    "tpw": 174,
+    "tr": 175,
+    "tt": 176,
+    "udm": 177,
+    "uig": 178,
+    "uk": 179,
+    "ur": 180,
+    "uz": 181,
+    "vec": 182,
+    "vi": 183,
+    "war": 184,
+    "wo": 185,
+    "wuu": 186,
+    "xal": 187,
+    "xh": 188,
+    "xmf": 189,
+    "yi": 190,
+    "yue": 191,
+    "zgh": 192,
+    "zh-Hant": 193,
+    "zlm": 194,
+    "zza": 195
   },
   "layer_norm_eps": 1e-05,
   "max_position_embeddings": 514,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3c40ea45cdd4c1efc24e5836eda25077176eeb2e7a7f6e93148e351ab63eb88f
-size 1112848117
+oid sha256:efa75cbb3caf556a4a25b82e1a16e48631200f49764d3d3fea65d055d0763496
+size 1112850933