aminnaghavi commited on
Commit
4c205ca
1 Parent(s): 97d3d0c

add tokenizer

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