Cantaosu commited on
Commit
bf58cf8
1 Parent(s): 468e806

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +38 -38
vocab.json CHANGED
@@ -1,43 +1,43 @@
1
  {
2
- "$": 37,
3
- "'": 11,
4
- "0": 1,
5
- "1": 20,
6
- "2": 33,
7
- "3": 12,
8
- "4": 34,
9
- "5": 21,
10
- "6": 30,
11
- "7": 2,
12
- "8": 23,
13
- "9": 31,
14
  "[PAD]": 40,
15
  "[UNK]": 39,
16
- "a": 10,
17
- "b": 24,
18
- "c": 38,
19
- "d": 15,
20
- "e": 25,
21
- "f": 18,
22
- "g": 35,
23
- "h": 6,
24
- "i": 19,
25
- "j": 17,
26
- "k": 26,
27
- "l": 8,
28
- "m": 0,
29
- "n": 3,
30
- "o": 4,
31
  "p": 27,
32
- "q": 28,
33
- "r": 32,
34
- "s": 16,
35
- "t": 9,
36
- "u": 29,
37
- "v": 13,
38
- "w": 14,
39
- "x": 7,
40
- "y": 36,
41
- "z": 22,
42
- "|": 5
43
  }
 
1
  {
2
+ "$": 21,
3
+ "'": 12,
4
+ "0": 37,
5
+ "1": 16,
6
+ "2": 32,
7
+ "3": 14,
8
+ "4": 18,
9
+ "5": 24,
10
+ "6": 0,
11
+ "7": 38,
12
+ "8": 30,
13
+ "9": 3,
14
  "[PAD]": 40,
15
  "[UNK]": 39,
16
+ "a": 29,
17
+ "b": 6,
18
+ "c": 31,
19
+ "d": 9,
20
+ "e": 34,
21
+ "f": 35,
22
+ "g": 26,
23
+ "h": 4,
24
+ "i": 8,
25
+ "j": 25,
26
+ "k": 2,
27
+ "l": 15,
28
+ "m": 11,
29
+ "n": 20,
30
+ "o": 7,
31
  "p": 27,
32
+ "q": 19,
33
+ "r": 10,
34
+ "s": 22,
35
+ "t": 36,
36
+ "u": 23,
37
+ "v": 5,
38
+ "w": 17,
39
+ "x": 28,
40
+ "y": 33,
41
+ "z": 13,
42
+ "|": 1
43
  }