YassineBenlaria commited on
Commit
c14f2dc
1 Parent(s): 6a41e95

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +35 -35
vocab.json CHANGED
@@ -1,39 +1,39 @@
1
  {
2
- "2": 10,
3
  "[PAD]": 36,
4
  "[UNK]": 35,
5
- "|": 2,
6
- "ء": 21,
7
- "ئ": 5,
8
- "ا": 0,
9
- "ب": 23,
10
- "ة": 8,
11
- "ت": 6,
12
- "ث": 14,
13
- "ج": 32,
14
- "ح": 33,
15
- "خ": 20,
16
- "د": 17,
17
- "ذ": 30,
18
- "ر": 1,
19
- "ز": 25,
20
- "س": 22,
21
- "ش": 24,
22
- "ص": 4,
23
- "ض": 15,
24
- "ط": 12,
25
- "ظ": 29,
26
- "ع": 3,
27
- "غ": 28,
28
- "ف": 9,
29
- "ق": 26,
30
- "ك": 31,
31
- "ل": 16,
32
- "م": 27,
33
- "ن": 7,
34
- "ه": 11,
35
- "و": 18,
36
- "ى": 13,
37
- "ي": 34,
38
- "ڤ": 19
39
  }
 
1
  {
2
+ "2": 25,
3
  "[PAD]": 36,
4
  "[UNK]": 35,
5
+ "|": 30,
6
+ "ء": 5,
7
+ "ئ": 21,
8
+ "ا": 28,
9
+ "ب": 26,
10
+ "ة": 24,
11
+ "ت": 16,
12
+ "ث": 31,
13
+ "ج": 17,
14
+ "ح": 10,
15
+ "خ": 3,
16
+ "د": 7,
17
+ "ذ": 6,
18
+ "ر": 2,
19
+ "ز": 9,
20
+ "س": 4,
21
+ "ش": 22,
22
+ "ص": 33,
23
+ "ض": 20,
24
+ "ط": 32,
25
+ "ظ": 1,
26
+ "ع": 34,
27
+ "غ": 18,
28
+ "ف": 14,
29
+ "ق": 15,
30
+ "ك": 8,
31
+ "ل": 19,
32
+ "م": 13,
33
+ "ن": 0,
34
+ "ه": 29,
35
+ "و": 12,
36
+ "ى": 11,
37
+ "ي": 23,
38
+ "ڤ": 27
39
  }