samuelchiji commited on
Commit
60e316f
1 Parent(s): 2e6cc51

Upload tokenizer

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