Upload tokenizer
Browse files- tokenizer.json +49 -1
- vocab.json +1 -1
tokenizer.json
CHANGED
@@ -106,7 +106,55 @@
|
|
106 |
"PAD": 1,
|
107 |
"WORD_BOUNDARY": 2,
|
108 |
"UTT_BOUNDARY": 3,
|
109 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
110 |
},
|
111 |
"unk_token": "UNK"
|
112 |
}
|
|
|
106 |
"PAD": 1,
|
107 |
"WORD_BOUNDARY": 2,
|
108 |
"UTT_BOUNDARY": 3,
|
109 |
+
"m": 4,
|
110 |
+
"ɛ": 5,
|
111 |
+
"ɲ": 6,
|
112 |
+
"y": 7,
|
113 |
+
"n": 8,
|
114 |
+
"k": 9,
|
115 |
+
"ɑ": 10,
|
116 |
+
"r": 11,
|
117 |
+
"aː": 12,
|
118 |
+
"d": 13,
|
119 |
+
"i": 14,
|
120 |
+
"o": 15,
|
121 |
+
"h": 16,
|
122 |
+
"z": 17,
|
123 |
+
"v": 18,
|
124 |
+
"l": 19,
|
125 |
+
"eː": 20,
|
126 |
+
"j": 21,
|
127 |
+
"ʃ": 22,
|
128 |
+
"ɟ": 23,
|
129 |
+
"s": 24,
|
130 |
+
"oː": 25,
|
131 |
+
"p": 26,
|
132 |
+
"t": 27,
|
133 |
+
"tsː": 28,
|
134 |
+
"b": 29,
|
135 |
+
"u": 30,
|
136 |
+
"ɡ": 31,
|
137 |
+
"tː": 32,
|
138 |
+
"f": 33,
|
139 |
+
"ø": 34,
|
140 |
+
"t̠ʃ": 35,
|
141 |
+
"uː": 36,
|
142 |
+
"iː": 37,
|
143 |
+
"ts": 38,
|
144 |
+
"ɟː": 39,
|
145 |
+
"yː": 40,
|
146 |
+
"øː": 41,
|
147 |
+
"ʎ": 42,
|
148 |
+
"t̠ʃː": 43,
|
149 |
+
"c": 44,
|
150 |
+
"ɡː": 45,
|
151 |
+
"kː": 46,
|
152 |
+
"ɑː": 47,
|
153 |
+
"dː": 48,
|
154 |
+
"pː": 49,
|
155 |
+
"ʒ": 50,
|
156 |
+
"cː": 51,
|
157 |
+
"bː": 52
|
158 |
},
|
159 |
"unk_token": "UNK"
|
160 |
}
|
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,"m":4,"ɛ":5,"ɲ":6,"y":7,"n":8,"k":9,"ɑ":10,"r":11,"aː":12,"d":13,"i":14,"o":15,"h":16,"z":17,"v":18,"l":19,"eː":20,"j":21,"ʃ":22,"ɟ":23,"s":24,"oː":25,"p":26,"t":27,"tsː":28,"b":29,"u":30,"ɡ":31,"tː":32,"f":33,"ø":34,"t̠ʃ":35,"uː":36,"iː":37,"ts":38,"ɟː":39,"yː":40,"øː":41,"ʎ":42,"t̠ʃː":43,"c":44,"ɡː":45,"kː":46,"ɑː":47,"dː":48,"pː":49,"ʒ":50,"cː":51,"bː":52}
|