akhikhan123 commited on
Commit
661cb62
·
verified ·
1 Parent(s): b376f8e

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +130 -130
vocab.json CHANGED
@@ -1,137 +1,137 @@
1
  {
2
- "!": 52,
3
  "\"": 128,
4
- "&": 45,
5
- "'": 46,
6
- "(": 82,
7
- ")": 91,
8
- ",": 59,
9
- "-": 48,
10
- ".": 23,
11
- "/": 78,
12
- ":": 105,
13
- ";": 61,
14
- "=": 43,
15
- "?": 64,
16
- "A": 84,
17
- "B": 83,
18
- "C": 107,
19
- "D": 55,
20
- "E": 37,
21
- "F": 26,
22
- "G": 79,
23
- "H": 122,
24
- "I": 76,
25
- "J": 18,
26
- "K": 32,
27
- "L": 39,
28
- "M": 99,
29
- "N": 11,
30
  "O": 86,
31
- "P": 35,
32
- "Q": 118,
33
- "R": 27,
34
- "S": 1,
35
- "T": 33,
36
- "U": 127,
37
- "V": 31,
38
- "W": 56,
39
- "X": 90,
40
- "Y": 53,
41
- "Z": 25,
42
- "[": 51,
43
  "[PAD]": 134,
44
  "[UNK]": 133,
45
- "]": 124,
46
- "a": 110,
47
- "b": 125,
48
- "c": 10,
49
- "d": 121,
50
- "e": 129,
51
- "f": 66,
52
  "g": 114,
53
- "h": 126,
54
- "i": 21,
55
- "j": 41,
56
- "k": 38,
57
- "l": 109,
58
- "m": 16,
59
- "n": 102,
60
- "o": 68,
61
- "p": 24,
62
- "q": 2,
63
- "r": 106,
64
- "s": 12,
65
- "t": 101,
66
- "u": 87,
67
- "v": 29,
68
- "w": 108,
69
- "x": 73,
70
- "y": 13,
71
- "z": 0,
72
- "|": 22,
73
- "¡": 115,
74
- "§": 94,
75
- "«": 6,
76
- "´": 49,
77
- "»": 62,
78
- "Á": 116,
79
- "Ã": 17,
80
- "È": 96,
81
- "É": 14,
82
- "Ö": 20,
83
- "Ú": 34,
84
- "ß": 36,
85
- "à": 7,
86
- "á": 65,
87
- "â": 113,
88
- "ã": 112,
89
- "ä": 9,
90
- "æ": 80,
91
- "ç": 77,
92
- "è": 47,
93
- "é": 50,
94
- "ê": 42,
95
- "ë": 74,
96
- "í": 40,
97
- "ï": 103,
98
- "ð": 95,
99
- "ñ": 5,
100
- "ò": 92,
101
- "ó": 132,
102
- "ô": 81,
103
- "ö": 69,
104
- "ø": 98,
105
- "ú": 75,
106
- "û": 131,
107
- "ü": 15,
108
- "þ": 19,
109
- "ā": 117,
110
- "ć": 88,
111
- "č": 70,
112
- "ł": 44,
113
- "ň": 58,
114
- "ō": 89,
115
- "Š": 93,
116
- "ū": 3,
117
- "π": 85,
118
- "В": 100,
119
- "е": 57,
120
- "з": 120,
121
- "и": 67,
122
- "й": 8,
123
- "к": 72,
124
- "л": 71,
125
- "н": 130,
126
- "ь": 30,
127
- "я": 104,
128
- "ụ": 119,
129
- "–": 111,
130
- "—": 60,
131
- "‘": 123,
132
- "’": 63,
133
- "“": 97,
134
- "”": 4,
135
- "…": 28,
136
- "€": 54
137
  }
 
1
  {
2
+ "!": 5,
3
  "\"": 128,
4
+ "&": 81,
5
+ "'": 56,
6
+ "(": 126,
7
+ ")": 118,
8
+ ",": 85,
9
+ "-": 132,
10
+ ".": 78,
11
+ "/": 25,
12
+ ":": 51,
13
+ ";": 68,
14
+ "=": 44,
15
+ "?": 75,
16
+ "A": 20,
17
+ "B": 24,
18
+ "C": 39,
19
+ "D": 16,
20
+ "E": 4,
21
+ "F": 72,
22
+ "G": 29,
23
+ "H": 55,
24
+ "I": 87,
25
+ "J": 90,
26
+ "K": 73,
27
+ "L": 119,
28
+ "M": 131,
29
+ "N": 122,
30
  "O": 86,
31
+ "P": 106,
32
+ "Q": 28,
33
+ "R": 48,
34
+ "S": 53,
35
+ "T": 97,
36
+ "U": 12,
37
+ "V": 77,
38
+ "W": 96,
39
+ "X": 120,
40
+ "Y": 99,
41
+ "Z": 47,
42
+ "[": 41,
43
  "[PAD]": 134,
44
  "[UNK]": 133,
45
+ "]": 100,
46
+ "a": 23,
47
+ "b": 50,
48
+ "c": 123,
49
+ "d": 35,
50
+ "e": 7,
51
+ "f": 40,
52
  "g": 114,
53
+ "h": 74,
54
+ "i": 93,
55
+ "j": 33,
56
+ "k": 57,
57
+ "l": 89,
58
+ "m": 0,
59
+ "n": 64,
60
+ "o": 46,
61
+ "p": 70,
62
+ "q": 107,
63
+ "r": 66,
64
+ "s": 62,
65
+ "t": 52,
66
+ "u": 69,
67
+ "v": 111,
68
+ "w": 124,
69
+ "x": 1,
70
+ "y": 115,
71
+ "z": 45,
72
+ "|": 34,
73
+ "¡": 19,
74
+ "§": 36,
75
+ "«": 21,
76
+ "´": 30,
77
+ "»": 2,
78
+ "Á": 104,
79
+ "Ã": 31,
80
+ "È": 8,
81
+ "É": 76,
82
+ "Ö": 127,
83
+ "Ú": 121,
84
+ "ß": 17,
85
+ "à": 49,
86
+ "á": 26,
87
+ "â": 109,
88
+ "ã": 108,
89
+ "ä": 15,
90
+ "æ": 58,
91
+ "ç": 13,
92
+ "è": 79,
93
+ "é": 9,
94
+ "ê": 113,
95
+ "ë": 83,
96
+ "í": 38,
97
+ "ï": 95,
98
+ "ð": 117,
99
+ "ñ": 80,
100
+ "ò": 11,
101
+ "ó": 103,
102
+ "ô": 54,
103
+ "ö": 59,
104
+ "ø": 37,
105
+ "ú": 91,
106
+ "û": 71,
107
+ "ü": 125,
108
+ "þ": 63,
109
+ "ā": 129,
110
+ "ć": 10,
111
+ "č": 82,
112
+ "ł": 27,
113
+ "ň": 110,
114
+ "ō": 105,
115
+ "Š": 43,
116
+ "ū": 130,
117
+ "π": 60,
118
+ "В": 84,
119
+ "е": 22,
120
+ "з": 94,
121
+ "и": 101,
122
+ "й": 67,
123
+ "к": 14,
124
+ "л": 116,
125
+ "н": 42,
126
+ "ь": 6,
127
+ "я": 112,
128
+ "ụ": 98,
129
+ "–": 88,
130
+ "—": 65,
131
+ "‘": 61,
132
+ "’": 18,
133
+ "“": 102,
134
+ "”": 92,
135
+ "…": 32,
136
+ "€": 3
137
  }