Martha-987 commited on
Commit
0a6a554
1 Parent(s): c8f657e

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +60 -27
vocab.json CHANGED
@@ -1,32 +1,65 @@
1
  {
 
 
2
  "'": 9,
3
- "[PAD]": 29,
4
- "[UNK]": 28,
5
- "a": 10,
6
- "b": 22,
7
- "c": 4,
8
- "d": 23,
9
- "e": 13,
10
- "f": 1,
11
- "g": 20,
12
- "h": 16,
13
- "i": 14,
14
- "j": 24,
15
- "k": 5,
16
- "l": 11,
17
- "m": 7,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
18
  "n": 21,
19
- "o": 26,
20
- "p": 6,
21
- "q": 3,
22
- "r": 2,
23
- "s": 0,
24
- "t": 19,
25
- "u": 15,
26
  "v": 8,
27
- "w": 18,
28
- "x": 12,
29
- "y": 25,
30
- "z": 17,
31
- "|": 27
32
  }
 
1
  {
2
+ "!": 59,
3
+ "\"": 14,
4
  "'": 9,
5
+ ",": 56,
6
+ "-": 5,
7
+ ".": 54,
8
+ ":": 47,
9
+ ";": 4,
10
+ "?": 32,
11
+ "A": 40,
12
+ "B": 3,
13
+ "C": 26,
14
+ "D": 46,
15
+ "E": 19,
16
+ "F": 20,
17
+ "G": 11,
18
+ "H": 39,
19
+ "I": 34,
20
+ "J": 18,
21
+ "K": 30,
22
+ "L": 31,
23
+ "M": 23,
24
+ "N": 38,
25
+ "O": 13,
26
+ "P": 10,
27
+ "Q": 52,
28
+ "R": 0,
29
+ "S": 7,
30
+ "T": 25,
31
+ "U": 44,
32
+ "V": 22,
33
+ "W": 37,
34
+ "X": 27,
35
+ "Y": 12,
36
+ "[PAD]": 62,
37
+ "[UNK]": 61,
38
+ "a": 41,
39
+ "b": 53,
40
+ "c": 6,
41
+ "d": 55,
42
+ "e": 45,
43
+ "f": 29,
44
+ "g": 17,
45
+ "h": 49,
46
+ "i": 15,
47
+ "j": 57,
48
+ "k": 33,
49
+ "l": 42,
50
+ "m": 36,
51
  "n": 21,
52
+ "o": 58,
53
+ "p": 35,
54
+ "q": 2,
55
+ "r": 1,
56
+ "s": 28,
57
+ "t": 51,
58
+ "u": 48,
59
  "v": 8,
60
+ "w": 16,
61
+ "x": 43,
62
+ "y": 24,
63
+ "z": 50,
64
+ "|": 60
65
  }