khaoulaoub commited on
Commit
88b2c43
1 Parent(s): d2474bb

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +33 -33
vocab.json CHANGED
@@ -1,37 +1,37 @@
1
  {
2
  "[PAD]": 34,
3
  "[UNK]": 33,
4
- "|": 23,
5
- "ء": 22,
6
- "ا": 4,
7
- "ب": 2,
8
- "ة": 16,
9
- "ت": 3,
10
- "ث": 25,
11
- "ج": 7,
12
- "ح": 1,
13
- "خ": 21,
14
- "د": 12,
15
- "ذ": 17,
16
- "ر": 0,
17
- "ز": 15,
18
- "س": 31,
19
- "ش": 11,
20
- "ص": 30,
21
- "ض": 6,
22
- "ط": 26,
23
- "ظ": 5,
24
- "ع": 18,
25
- "غ": 29,
26
- "ـ": 19,
27
- "ف": 14,
28
- "ق": 10,
29
- "ك": 8,
30
- "ل": 32,
31
- "م": 27,
32
- "ن": 9,
33
- "ه": 28,
34
- "و": 24,
35
- "ى": 13,
36
- "ي": 20
37
  }
 
1
  {
2
  "[PAD]": 34,
3
  "[UNK]": 33,
4
+ "|": 1,
5
+ "ء": 25,
6
+ "ا": 8,
7
+ "ب": 15,
8
+ "ة": 18,
9
+ "ت": 12,
10
+ "ث": 0,
11
+ "ج": 28,
12
+ "ح": 17,
13
+ "خ": 22,
14
+ "د": 14,
15
+ "ذ": 4,
16
+ "ر": 26,
17
+ "ز": 21,
18
+ "س": 24,
19
+ "ش": 19,
20
+ "ص": 9,
21
+ "ض": 23,
22
+ "ط": 32,
23
+ "ظ": 3,
24
+ "ع": 11,
25
+ "غ": 5,
26
+ "ـ": 13,
27
+ "ف": 2,
28
+ "ق": 6,
29
+ "ك": 20,
30
+ "ل": 10,
31
+ "م": 31,
32
+ "ن": 30,
33
+ "ه": 27,
34
+ "و": 29,
35
+ "ى": 7,
36
+ "ي": 16
37
  }