mgtoxd commited on
Commit
0e761fc
·
1 Parent(s): 21b55b1

add tokenizer

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