Upload tokenizer

tokenizer.json

@@ -136,233 +136,117 @@
       "PAD": 1,
       "W": 2,
       "UTT_BOUNDARY": 3,
-      "d": 4,
-      "o": 5,
-      "y": 6,
-      "u": 7,
-      "w": 8,
-      "a": 9,
-      "n": 10,
-      "t": 11,
-      "l": 12,
-      "k": 13,
-      "h": 14,
-      "i": 15,
-      "s": 16,
-      "?": 17,
-      "e": 18,
-      ".": 19,
-      "r": 20,
-      "'": 21,
-      "f": 22,
-      "c": 23,
-      "g": 24,
-      "p": 25,
-      "b": 26,
-      "m": 27,
-      "v": 28,
-      "j": 29,
-      "!": 30,
-      "x": 31,
-      "q": 32,
-      "z": 33,
-      "-": 34,
-      "&": 35,
-      ",": 36,
-      "/": 37,
-      "1": 38,
-      "9": 39,
-      "5": 40,
-      "0": 41,
-      ";": 42,
-      "‘": 43,
-      "’": 44,
-      "—": 45,
-      ":": 46,
-      "+": 47,
-      "8": 48,
-      "3": 49,
-      "7": 50,
-      "4": 51,
-      "6": 52,
-      "2": 53,
-      "=": 54,
+      "y": 4,
+      "e": 5,
+      "a": 6,
+      "h": 7,
+      ".": 8,
+      "c": 9,
+      "o": 10,
+      "m": 11,
+      "p": 12,
+      "u": 13,
+      "n": 14,
+      "d": 15,
+      "'": 16,
+      "s": 17,
+      "t": 18,
+      "i": 19,
+      "g": 20,
+      "l": 21,
+      "k": 22,
+      "x": 23,
+      ",": 24,
+      "r": 25,
+      "w": 26,
+      "v": 27,
+      "f": 28,
+      "b": 29,
+      "j": 30,
+      "?": 31,
+      "-": 32,
+      "q": 33,
+      ";": 34,
+      "2": 35,
+      "‘": 36,
+      "’": 37,
+      "!": 38,
+      "/": 39,
+      "1": 40,
+      ":": 41,
+      "z": 42,
+      "3": 43,
+      "6": 44,
+      "9": 45,
+      "&": 46,
+      "4": 47,
+      "5": 48,
+      "0": 49,
+      "=": 50,
+      "8": 51,
+      "7": 52,
+      "£": 53,
+      "(": 54,
       ")": 55,
-      "(": 56,
-      "_": 57,
-      "*": 58,
-      "£": 59,
-      "–": 60,
-      "#": 61,
-      "`": 62,
-      "\"": 63,
-      "æ": 64,
-      "]": 65,
-      "|": 66,
-      "$": 67,
-      "“": 68,
-      "”": 69,
-      "[": 70,
-      "œ": 71,
-      "{": 72,
-      "}": 73,
-      "…": 74,
+      "—": 56,
+      "*": 57,
+      "]": 58,
+      "[": 59,
+      "\"": 60,
+      "_": 61,
+      "%": 62,
+      "“": 63,
+      "”": 64,
+      "+": 65,
+      "$": 66,
+      "^": 67,
+      "#": 68,
+      "æ": 69,
+      "ʌ": 70,
+      "ɩ": 71,
+      "ə": 72,
+      "↫": 73,
+      "|": 74,
       "°": 75,
-      "§": 76,
-      ">": 77,
-      "·": 78,
-      "¢": 79,
-      "%": 80,
-      "^": 81,
-      "½": 82,
-      "¶": 83,
-      "×": 84,
-      "¼": 85,
-      "¾": 86,
-      "φ": 87,
-      "<": 88,
-      "´": 89,
-      "¯": 90,
-      "¦": 91,
-      "†": 92,
-      "«": 93,
-      "¬": 94,
-      "©": 95,
-      "“": 96,
-      "●": 97,
-      "¹": 98,
-      "⁄": 99,
-      "²": 100,
-      "₂": 101,
-      "│": 102,
-      "δ": 103,
-      "α": 104,
-      "ο": 105,
-      "υ": 106,
-      "χ": 107,
-      "ι": 108,
-      "η": 109,
-      "μ": 110,
-      "ε": 111,
-      "γ": 112,
-      "λ": 113,
-      "β": 114,
-      "κ": 115,
-      "ω": 116,
-      "ν": 117,
-      "ρ": 118,
-      "σ": 119,
-      "ς": 120,
-      "τ": 121,
-      "θ": 122,
-      "π": 123,
-      "′": 124,
-      "³": 125,
-      "⸺": 126,
-      "―": 127,
-      "±": 128,
-      "~": 129,
-      "ß": 130,
-      "µ": 131,
-      "„": 132,
-      "@": 133,
-      "\\": 134,
-      "♪": 135,
-      "¿": 136,
-      "¡": 137,
-      "​": 138,
-      "─": 139,
-      "–": 140,
-      "ð": 141,
-      "­": 142,
-      "º": 143,
-      "¸": 144,
-      "þ": 145,
-      "и": 146,
-      "♫": 147,
-      "¤": 148,
-      "¨": 149,
-      "ø": 150,
-      "": 151,
-      "ª": 152,
-      "đ": 153,
-      "€": 154,
-      "™": 155,
-      "€": 156,
-      "fl": 157,
-      "♬": 158,
-      "’": 159,
-      "，": 160,
-      "": 161,
-      "™": 162,
-      "®": 163,
-      "ı": 164,
-      "с": 165,
-      "в": 166,
-      "‎": 167,
-      "ц": 168,
-      "ь": 169,
-      "д": 170,
-      "н": 171,
-      "е": 172,
-      "м": 173,
-      "о": 174,
-      "т": 175,
-      "г": 176,
-      "а": 177,
-      "р": 178,
-      "ч": 179,
-      "п": 180,
-      "я": 181,
-      "б": 182,
-      "•": 183,
-      "ł": 184,
-      "ร": 185,
-      "ก": 186,
-      "к": 187,
-      "у": 188,
-      "‚": 189,
-      "ˆ": 190,
-      "„": 191,
-      "ƒ": 192,
-      "‡": 193,
-      "Š": 194,
-      "‘": 195,
-      "大": 196,
-      "”": 197,
-      "ا": 198,
-      "л": 199,
-      "з": 200,
-      "‑": 201,
-      "ʻ": 202,
-      "ت": 203,
-      "ו": 204,
-      "י": 205,
-      "ر": 206,
-      "ي": 207,
-      "ل": 208,
-      "ه": 209,
-      "ع": 210,
-      "ن": 211,
-      "ə": 212,
-      "−": 213,
-      "→": 214,
-      "ы": 215,
-      "ː": 216,
-      "و": 217,
-      "م": 218,
-      "س": 219,
-      "ح": 220,
-      "د": 221,
-      "ب": 222,
-      "ی": 223,
-      "»": 224,
-      "น": 225,
-      "า": 226,
-      "ง": 227,
-      "ม": 228,
-      "ʼ": 229,
-      "ˈ": 230
+      "ø": 76,
+      "~": 77,
+      "⁄": 78,
+      "`": 79,
+      "�": 80,
+      "′": 81,
+      "@": 82,
+      "}": 83,
+      "{": 84,
+      "―": 85,
+      "–": 86,
+      "·": 87,
+      "♪": 88,
+      "¡": 89,
+      "÷": 90,
+      "\\": 91,
+      "¶": 92,
+      "ð": 93,
+      "¿": 94,
+      "­": 95,
+      "♫": 96,
+      "​": 97,
+      "œ": 98,
+      "ł": 99,
+      "¦": 100,
+      "×": 101,
+      "™": 102,
+      "ß": 103,
+      "ˈ": 104,
+      "ı": 105,
+      "đ": 106,
+      "−": 107,
+      "ː": 108,
+      "•": 109,
+      "⟨": 110,
+      "⟩": 111,
+      "ŋ": 112,
+      "ʼ": 113,
+      "\t": 114
     },
     "unk_token": "UNK"
   }

vocab.json

@@ -1 +1 @@
-{"UNK":0,"PAD":1,"W":2,"UTT_BOUNDARY":3,"d":4,"o":5,"y":6,"u":7,"w":8,"a":9,"n":10,"t":11,"l":12,"k":13,"h":14,"i":15,"s":16,"?":17,"e":18,".":19,"r":20,"'":21,"f":22,"c":23,"g":24,"p":25,"b":26,"m":27,"v":28,"j":29,"!":30,"x":31,"q":32,"z":33,"-":34,"&":35,",":36,"/":37,"1":38,"9":39,"5":40,"0":41,";":42,"‘":43,"’":44,"—":45,":":46,"+":47,"8":48,"3":49,"7":50,"4":51,"6":52,"2":53,"=":54,")":55,"(":56,"_":57,"*":58,"£":59,"–":60,"#":61,"`":62,"\"":63,"æ":64,"]":65,"|":66,"$":67,"“":68,"”":69,"[":70,"œ":71,"{":72,"}":73,"…":74,"°":75,"§":76,">":77,"·":78,"¢":79,"%":80,"^":81,"½":82,"¶":83,"×":84,"¼":85,"¾":86,"φ":87,"<":88,"´":89,"¯":90,"¦":91,"†":92,"«":93,"¬":94,"©":95,"“":96,"●":97,"¹":98,"⁄":99,"²":100,"₂":101,"│":102,"δ":103,"α":104,"ο":105,"υ":106,"χ":107,"ι":108,"η":109,"μ":110,"ε":111,"γ":112,"λ":113,"β":114,"κ":115,"ω":116,"ν":117,"ρ":118,"σ":119,"ς":120,"τ":121,"θ":122,"π":123,"′":124,"³":125,"⸺":126,"―":127,"±":128,"~":129,"ß":130,"µ":131,"„":132,"@":133,"\\":134,"♪":135,"¿":136,"¡":137,"​":138,"─":139,"–":140,"ð":141,"­":142,"º":143,"¸":144,"þ":145,"и":146,"♫":147,"¤":148,"¨":149,"ø":150,"":151,"ª":152,"đ":153,"€":154,"™":155,"€":156,"fl":157,"♬":158,"’":159,"，":160,"":161,"™":162,"®":163,"ı":164,"с":165,"в":166,"‎":167,"ц":168,"ь":169,"д":170,"н":171,"е":172,"м":173,"о":174,"т":175,"г":176,"а":177,"р":178,"ч":179,"п":180,"я":181,"б":182,"•":183,"ł":184,"ร":185,"ก":186,"к":187,"у":188,"‚":189,"ˆ":190,"„":191,"ƒ":192,"‡":193,"Š":194,"‘":195,"大":196,"”":197,"ا":198,"л":199,"з":200,"‑":201,"ʻ":202,"ت":203,"ו":204,"י":205,"ر":206,"ي":207,"ل":208,"ه":209,"ع":210,"ن":211,"ə":212,"−":213,"→":214,"ы":215,"ː":216,"و":217,"م":218,"س":219,"ح":220,"د":221,"ب":222,"ی":223,"»":224,"น":225,"า":226,"ง":227,"ม":228,"ʼ":229,"ˈ":230}
+{"UNK":0,"PAD":1,"W":2,"UTT_BOUNDARY":3,"y":4,"e":5,"a":6,"h":7,".":8,"c":9,"o":10,"m":11,"p":12,"u":13,"n":14,"d":15,"'":16,"s":17,"t":18,"i":19,"g":20,"l":21,"k":22,"x":23,",":24,"r":25,"w":26,"v":27,"f":28,"b":29,"j":30,"?":31,"-":32,"q":33,";":34,"2":35,"‘":36,"’":37,"!":38,"/":39,"1":40,":":41,"z":42,"3":43,"6":44,"9":45,"&":46,"4":47,"5":48,"0":49,"=":50,"8":51,"7":52,"£":53,"(":54,")":55,"—":56,"*":57,"]":58,"[":59,"\"":60,"_":61,"%":62,"“":63,"”":64,"+":65,"$":66,"^":67,"#":68,"æ":69,"ʌ":70,"ɩ":71,"ə":72,"↫":73,"|":74,"°":75,"ø":76,"~":77,"⁄":78,"`":79,"�":80,"′":81,"@":82,"}":83,"{":84,"―":85,"–":86,"·":87,"♪":88,"¡":89,"÷":90,"\\":91,"¶":92,"ð":93,"¿":94,"­":95,"♫":96,"​":97,"œ":98,"ł":99,"¦":100,"×":101,"™":102,"ß":103,"ˈ":104,"ı":105,"đ":106,"−":107,"ː":108,"•":109,"⟨":110,"⟩":111,"ŋ":112,"ʼ":113,"\t":114}