Thunpitcha commited on
Commit
368c9bb
1 Parent(s): d107d22

Upload tokenizer

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