nickdee96 commited on
Commit
c50daae
·
1 Parent(s): 4408f42

Upload tokenizer

Browse files
Files changed (2) hide show
  1. added_tokens.json +2 -2
  2. vocab.json +81 -31
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 46,
3
- "<s>": 45
4
  }
 
1
  {
2
+ "</s>": 96,
3
+ "<s>": 95
4
  }
vocab.json CHANGED
@@ -13,35 +13,85 @@
13
  ";": 12,
14
  "=": 13,
15
  "?": 14,
16
- "[PAD]": 44,
17
- "[UNK]": 43,
18
- "_": 15,
19
- "`": 16,
20
- "a": 17,
21
- "b": 18,
22
- "c": 19,
23
- "d": 20,
24
- "e": 21,
25
- "f": 22,
26
- "g": 23,
27
- "h": 24,
28
- "i": 25,
29
- "j": 26,
30
- "k": 27,
31
- "l": 28,
32
- "m": 29,
33
- "n": 30,
34
- "o": 31,
35
- "p": 32,
36
- "q": 33,
37
- "r": 34,
38
- "s": 35,
39
- "t": 36,
40
- "u": 37,
41
- "v": 38,
42
- "w": 39,
43
- "x": 40,
44
- "y": 41,
45
- "z": 42,
46
- "|": 0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
47
  }
 
13
  ";": 12,
14
  "=": 13,
15
  "?": 14,
16
+ "A": 15,
17
+ "B": 16,
18
+ "C": 17,
19
+ "D": 18,
20
+ "E": 19,
21
+ "F": 20,
22
+ "G": 21,
23
+ "H": 22,
24
+ "I": 23,
25
+ "J": 24,
26
+ "K": 25,
27
+ "L": 26,
28
+ "M": 27,
29
+ "N": 28,
30
+ "O": 29,
31
+ "P": 30,
32
+ "Q": 31,
33
+ "R": 32,
34
+ "S": 33,
35
+ "T": 34,
36
+ "U": 35,
37
+ "V": 36,
38
+ "W": 37,
39
+ "X": 38,
40
+ "Y": 39,
41
+ "Z": 40,
42
+ "[PAD]": 94,
43
+ "[UNK]": 93,
44
+ "_": 41,
45
+ "`": 42,
46
+ "a": 43,
47
+ "b": 44,
48
+ "c": 45,
49
+ "d": 46,
50
+ "e": 47,
51
+ "f": 48,
52
+ "g": 49,
53
+ "h": 50,
54
+ "i": 51,
55
+ "j": 52,
56
+ "k": 53,
57
+ "l": 54,
58
+ "m": 55,
59
+ "n": 56,
60
+ "o": 57,
61
+ "p": 58,
62
+ "q": 59,
63
+ "r": 60,
64
+ "s": 61,
65
+ "t": 62,
66
+ "u": 63,
67
+ "v": 64,
68
+ "w": 65,
69
+ "x": 66,
70
+ "y": 67,
71
+ "z": 68,
72
+ "|": 0,
73
+ "°": 69,
74
+ "µ": 70,
75
+ "Ã": 71,
76
+ "Å": 72,
77
+ "á": 73,
78
+ "â": 74,
79
+ "é": 75,
80
+ "ë": 76,
81
+ "í": 77,
82
+ "ï": 78,
83
+ "ñ": 79,
84
+ "ó": 80,
85
+ "ö": 81,
86
+ "ø": 82,
87
+ "ú": 83,
88
+ "Š": 84,
89
+ "ū": 85,
90
+ "ː": 86,
91
+ "ụ": 87,
92
+ "‘": 88,
93
+ "’": 89,
94
+ "”": 90,
95
+ "•": 91,
96
+ "…": 92
97
  }