samuelchiji commited on
Commit
fa93d56
1 Parent(s): 55f9c1f

Upload tokenizer

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