mgtoxd commited on
Commit
5e643b5
·
1 Parent(s): 27a91e6

add tokenizer

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