hbm209 commited on
Commit
89724e7
1 Parent(s): a10f38d

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +48 -48
vocab.json CHANGED
@@ -1,53 +1,53 @@
1
  {
2
  "[PAD]": 50,
3
  "[UNK]": 49,
4
- "a": 21,
5
- "b": 47,
6
- "c": 11,
7
- "e": 5,
8
- "g": 29,
9
- "h": 44,
10
- "i": 32,
11
- "k": 19,
12
- "l": 41,
13
- "m": 42,
14
- "n": 10,
15
- "o": 20,
16
- "p": 25,
17
- "q": 31,
18
- "r": 12,
19
- "s": 43,
20
- "t": 39,
21
- "u": 37,
22
  "v": 45,
23
- "x": 9,
24
- "y": 3,
25
- "|": 14,
26
- "à": 23,
27
- "á": 34,
28
- "ã": 26,
29
- "ê": 22,
30
- "ó": 0,
31
- "ô": 8,
32
- "ú": 4,
33
- "đ": 28,
34
- "ũ": 48,
35
- "ơ": 16,
36
- "ư": 18,
37
- "ạ": 15,
38
- "ả": 30,
39
- "ầ": 40,
40
- "ậ": 33,
41
- "ắ": 38,
42
- "ẹ": 7,
43
- "ẽ": 36,
44
- "ế": 2,
45
- "ề": 1,
46
- "ể": 17,
47
- "ọ": 35,
48
- "ố": 13,
49
- "ỗ": 46,
50
- "ớ": 6,
51
- "ờ": 24,
52
- "ợ": 27
53
  }
 
1
  {
2
  "[PAD]": 50,
3
  "[UNK]": 49,
4
+ "a": 39,
5
+ "b": 26,
6
+ "c": 4,
7
+ "e": 6,
8
+ "g": 41,
9
+ "h": 10,
10
+ "i": 34,
11
+ "k": 0,
12
+ "l": 31,
13
+ "m": 3,
14
+ "n": 29,
15
+ "o": 42,
16
+ "p": 17,
17
+ "q": 7,
18
+ "r": 19,
19
+ "s": 18,
20
+ "t": 48,
21
+ "u": 32,
22
  "v": 45,
23
+ "x": 14,
24
+ "y": 25,
25
+ "|": 24,
26
+ "à": 30,
27
+ "á": 16,
28
+ "ã": 12,
29
+ "ê": 33,
30
+ "ó": 38,
31
+ "ô": 28,
32
+ "ú": 22,
33
+ "đ": 1,
34
+ "ũ": 27,
35
+ "ơ": 20,
36
+ "ư": 2,
37
+ "ạ": 46,
38
+ "ả": 43,
39
+ "ầ": 21,
40
+ "ậ": 40,
41
+ "ắ": 23,
42
+ "ẹ": 37,
43
+ "ẽ": 35,
44
+ "ế": 44,
45
+ "ề": 15,
46
+ "ể": 11,
47
+ "ọ": 5,
48
+ "ố": 8,
49
+ "ỗ": 36,
50
+ "ớ": 9,
51
+ "ờ": 13,
52
+ "ợ": 47
53
  }