sanchit-gandhi HF staff commited on
Commit
4997cd1
1 Parent(s): 75c687b

Upload tokenizer

Browse files
Files changed (3) hide show
  1. special_tokens_map.json +4 -0
  2. tokenizer_config.json +12 -0
  3. vocab.json +179 -0
special_tokens_map.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "pad_token": "_",
3
+ "unk_token": "<unk>"
4
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_blank": true,
3
+ "clean_up_tokenization_spaces": true,
4
+ "is_uroman": false,
5
+ "language": null,
6
+ "model_max_length": 1000000000000000019884624838656,
7
+ "normalize": true,
8
+ "pad_token": "_",
9
+ "phonemize": true,
10
+ "tokenizer_class": "VitsTokenizer",
11
+ "unk_token": "<unk>"
12
+ }
vocab.json ADDED
@@ -0,0 +1,179 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ " ": 16,
3
+ "!": 5,
4
+ "\"": 11,
5
+ "'": 176,
6
+ ",": 3,
7
+ ".": 4,
8
+ ":": 2,
9
+ ";": 1,
10
+ "?": 6,
11
+ "A": 17,
12
+ "B": 18,
13
+ "C": 19,
14
+ "D": 20,
15
+ "E": 21,
16
+ "F": 22,
17
+ "G": 23,
18
+ "H": 24,
19
+ "I": 25,
20
+ "J": 26,
21
+ "K": 27,
22
+ "L": 28,
23
+ "M": 29,
24
+ "N": 30,
25
+ "O": 31,
26
+ "P": 32,
27
+ "Q": 33,
28
+ "R": 34,
29
+ "S": 35,
30
+ "T": 36,
31
+ "U": 37,
32
+ "V": 38,
33
+ "W": 39,
34
+ "X": 40,
35
+ "Y": 41,
36
+ "Z": 42,
37
+ "_": 0,
38
+ "a": 43,
39
+ "b": 44,
40
+ "c": 45,
41
+ "d": 46,
42
+ "e": 47,
43
+ "f": 48,
44
+ "g": 49,
45
+ "h": 50,
46
+ "i": 51,
47
+ "j": 52,
48
+ "k": 53,
49
+ "l": 54,
50
+ "m": 55,
51
+ "n": 56,
52
+ "o": 57,
53
+ "p": 58,
54
+ "q": 59,
55
+ "r": 60,
56
+ "s": 61,
57
+ "t": 62,
58
+ "u": 63,
59
+ "v": 64,
60
+ "w": 65,
61
+ "x": 66,
62
+ "y": 67,
63
+ "z": 68,
64
+ "¡": 7,
65
+ "«": 12,
66
+ "»": 13,
67
+ "¿": 8,
68
+ "æ": 72,
69
+ "ç": 78,
70
+ "ð": 81,
71
+ "ø": 116,
72
+ "ħ": 98,
73
+ "ŋ": 112,
74
+ "œ": 120,
75
+ "ǀ": 152,
76
+ "ǁ": 153,
77
+ "ǂ": 154,
78
+ "ǃ": 155,
79
+ "ɐ": 70,
80
+ "ɑ": 69,
81
+ "ɒ": 71,
82
+ "ɓ": 73,
83
+ "ɔ": 76,
84
+ "ɕ": 77,
85
+ "ɖ": 80,
86
+ "ɗ": 79,
87
+ "ɘ": 84,
88
+ "ə": 83,
89
+ "ɚ": 85,
90
+ "ɛ": 86,
91
+ "ɜ": 87,
92
+ "ɝ": 88,
93
+ "ɞ": 89,
94
+ "ɟ": 90,
95
+ "ɠ": 93,
96
+ "ɡ": 92,
97
+ "ɢ": 94,
98
+ "ɣ": 139,
99
+ "ɤ": 140,
100
+ "ɥ": 99,
101
+ "ɦ": 96,
102
+ "ɧ": 97,
103
+ "ɨ": 101,
104
+ "ɪ": 102,
105
+ "ɫ": 106,
106
+ "ɬ": 105,
107
+ "ɭ": 104,
108
+ "ɮ": 107,
109
+ "ɯ": 110,
110
+ "ɰ": 111,
111
+ "ɱ": 109,
112
+ "ɲ": 114,
113
+ "ɳ": 113,
114
+ "ɴ": 115,
115
+ "ɵ": 117,
116
+ "ɶ": 121,
117
+ "ɸ": 118,
118
+ "ɹ": 123,
119
+ "ɺ": 124,
120
+ "ɻ": 126,
121
+ "ɽ": 129,
122
+ "ɾ": 125,
123
+ "ʀ": 127,
124
+ "ʁ": 128,
125
+ "ʂ": 130,
126
+ "ʃ": 131,
127
+ "ʄ": 91,
128
+ "ʈ": 132,
129
+ "ʉ": 134,
130
+ "ʊ": 135,
131
+ "ʋ": 136,
132
+ "ʌ": 138,
133
+ "ʍ": 141,
134
+ "ʎ": 143,
135
+ "ʏ": 144,
136
+ "ʐ": 146,
137
+ "ʑ": 145,
138
+ "ʒ": 147,
139
+ "ʔ": 148,
140
+ "ʕ": 150,
141
+ "ʘ": 122,
142
+ "ʙ": 74,
143
+ "ʛ": 95,
144
+ "ʜ": 100,
145
+ "ʝ": 103,
146
+ "ʟ": 108,
147
+ "ʡ": 149,
148
+ "ʢ": 151,
149
+ "ʤ": 82,
150
+ "ʧ": 133,
151
+ "ʰ": 162,
152
+ "ʱ": 163,
153
+ "ʲ": 164,
154
+ "ʴ": 161,
155
+ "ʷ": 165,
156
+ "ʼ": 160,
157
+ "ˈ": 156,
158
+ "ˌ": 157,
159
+ "ː": 158,
160
+ "ˑ": 159,
161
+ "˞": 168,
162
+ "ˠ": 166,
163
+ "ˤ": 167,
164
+ "̩": 175,
165
+ "β": 75,
166
+ "θ": 119,
167
+ "χ": 142,
168
+ "ᵻ": 177,
169
+ "—": 9,
170
+ "“": 14,
171
+ "”": 15,
172
+ "…": 10,
173
+ "↑": 170,
174
+ "→": 171,
175
+ "↓": 169,
176
+ "↗": 172,
177
+ "↘": 173,
178
+ "ⱱ": 137
179
+ }