Upload tokenizer
Browse files- tokenizer.json +49 -1
- vocab.json +1 -1
tokenizer.json
CHANGED
@@ -105,7 +105,55 @@
|
|
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 |
+
"aː": 4,
|
110 |
+
"ɾ": 5,
|
111 |
+
"r": 6,
|
112 |
+
"ɪ": 7,
|
113 |
+
"ɛ": 8,
|
114 |
+
"d": 9,
|
115 |
+
"s": 10,
|
116 |
+
"j": 11,
|
117 |
+
"a": 12,
|
118 |
+
"b": 13,
|
119 |
+
"iː": 14,
|
120 |
+
"k": 15,
|
121 |
+
"ʋ": 16,
|
122 |
+
"ɛː": 17,
|
123 |
+
"θ": 18,
|
124 |
+
"i": 19,
|
125 |
+
"l": 20,
|
126 |
+
"n": 21,
|
127 |
+
"uː": 22,
|
128 |
+
"ð": 23,
|
129 |
+
"ɡ": 24,
|
130 |
+
"ɔ": 25,
|
131 |
+
"h": 26,
|
132 |
+
"aʊ": 27,
|
133 |
+
"y": 28,
|
134 |
+
"m": 29,
|
135 |
+
"f": 30,
|
136 |
+
"ɔː": 31,
|
137 |
+
"x": 32,
|
138 |
+
"ɟ": 33,
|
139 |
+
"t": 34,
|
140 |
+
"eɪ": 35,
|
141 |
+
"oʊ": 36,
|
142 |
+
"p": 37,
|
143 |
+
"ŋ": 38,
|
144 |
+
"ɣ": 39,
|
145 |
+
"yː": 40,
|
146 |
+
"u": 41,
|
147 |
+
"ɪː": 42,
|
148 |
+
"œ": 43,
|
149 |
+
"aɪ": 44,
|
150 |
+
"ç": 45,
|
151 |
+
"ə": 46,
|
152 |
+
"øy": 47,
|
153 |
+
"c": 48,
|
154 |
+
"ɲ": 49,
|
155 |
+
"œː": 50,
|
156 |
+
"ɔɪ": 51
|
157 |
},
|
158 |
"unk_token": "UNK"
|
159 |
}
|
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,"aː":4,"ɾ":5,"r":6,"ɪ":7,"ɛ":8,"d":9,"s":10,"j":11,"a":12,"b":13,"iː":14,"k":15,"ʋ":16,"ɛː":17,"θ":18,"i":19,"l":20,"n":21,"uː":22,"ð":23,"ɡ":24,"ɔ":25,"h":26,"aʊ":27,"y":28,"m":29,"f":30,"ɔː":31,"x":32,"ɟ":33,"t":34,"eɪ":35,"oʊ":36,"p":37,"ŋ":38,"ɣ":39,"yː":40,"u":41,"ɪː":42,"œ":43,"aɪ":44,"ç":45,"ə":46,"øy":47,"c":48,"ɲ":49,"œː":50,"ɔɪ":51}
|