Upload tokenizer
Browse files- tokenizer.json +42 -1
- vocab.json +1 -1
tokenizer.json
CHANGED
|
@@ -105,7 +105,48 @@
|
|
| 105 |
"UNK": 0,
|
| 106 |
"PAD": 1,
|
| 107 |
"WORD_BOUNDARY": 2,
|
| 108 |
-
"UTT_BOUNDARY": 3
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 109 |
},
|
| 110 |
"unk_token": "UNK"
|
| 111 |
}
|
|
|
|
| 105 |
"UNK": 0,
|
| 106 |
"PAD": 1,
|
| 107 |
"WORD_BOUNDARY": 2,
|
| 108 |
+
"UTT_BOUNDARY": 3,
|
| 109 |
+
"d": 4,
|
| 110 |
+
"e": 5,
|
| 111 |
+
"ɛ": 6,
|
| 112 |
+
"n": 7,
|
| 113 |
+
"m": 8,
|
| 114 |
+
"s": 9,
|
| 115 |
+
"t": 10,
|
| 116 |
+
"k": 11,
|
| 117 |
+
"j": 12,
|
| 118 |
+
"f": 13,
|
| 119 |
+
"ɑ": 14,
|
| 120 |
+
"ɒ": 15,
|
| 121 |
+
"ə": 16,
|
| 122 |
+
"ʋ": 17,
|
| 123 |
+
"a": 18,
|
| 124 |
+
"l": 19,
|
| 125 |
+
"h": 20,
|
| 126 |
+
"b": 21,
|
| 127 |
+
"ʁ": 22,
|
| 128 |
+
"p": 23,
|
| 129 |
+
"œ": 24,
|
| 130 |
+
"i": 25,
|
| 131 |
+
"ɡ": 26,
|
| 132 |
+
"ʌ": 27,
|
| 133 |
+
"u": 28,
|
| 134 |
+
"ʃ": 29,
|
| 135 |
+
"ɔ": 30,
|
| 136 |
+
"w": 31,
|
| 137 |
+
"ð": 32,
|
| 138 |
+
"o": 33,
|
| 139 |
+
"y": 34,
|
| 140 |
+
"ŋ": 35,
|
| 141 |
+
"aɪ": 36,
|
| 142 |
+
"œː": 37,
|
| 143 |
+
"aː": 38,
|
| 144 |
+
"d̠ʒ": 39,
|
| 145 |
+
"uː": 40,
|
| 146 |
+
"ʌː": 41,
|
| 147 |
+
"ɜ": 42,
|
| 148 |
+
"oː": 43,
|
| 149 |
+
"yː": 44
|
| 150 |
},
|
| 151 |
"unk_token": "UNK"
|
| 152 |
}
|
vocab.json
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
{"UNK":0,"PAD":1,"WORD_BOUNDARY":2,"UTT_BOUNDARY":3}
|
|
|
|
| 1 |
+
{"UNK":0,"PAD":1,"WORD_BOUNDARY":2,"UTT_BOUNDARY":3,"d":4,"e":5,"ɛ":6,"n":7,"m":8,"s":9,"t":10,"k":11,"j":12,"f":13,"ɑ":14,"ɒ":15,"ə":16,"ʋ":17,"a":18,"l":19,"h":20,"b":21,"ʁ":22,"p":23,"œ":24,"i":25,"ɡ":26,"ʌ":27,"u":28,"ʃ":29,"ɔ":30,"w":31,"ð":32,"o":33,"y":34,"ŋ":35,"aɪ":36,"œː":37,"aː":38,"d̠ʒ":39,"uː":40,"ʌː":41,"ɜ":42,"oː":43,"yː":44}
|