aminnaghavi commited on
Commit
47142cc
1 Parent(s): 81096eb

add tokenizer

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