YassineBenlaria commited on
Commit
ac38406
1 Parent(s): d0765fb

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +43 -45
vocab.json CHANGED
@@ -1,47 +1,45 @@
1
  {
2
- "[PAD]": 44,
3
- "[UNK]": 43,
4
- "|": 7,
5
- "ء": 2,
6
- "آ": 18,
7
- "أ": 35,
8
- "ؤ": 31,
9
- "إ": 23,
10
- "ئ": 42,
11
- "ا": 33,
12
- "ب": 3,
13
- "ة": 15,
14
- "ت": 29,
15
- "ث": 34,
16
- "ج": 9,
17
- "ح": 17,
18
- "خ": 37,
19
- "د": 40,
20
- "ذ": 12,
21
- "ر": 8,
22
- "ز": 6,
23
- "س": 10,
24
- "ش": 38,
25
- "ص": 20,
26
- "ض": 0,
27
- "ط": 32,
28
- "ظ": 14,
29
- "ع": 19,
30
- "غ": 21,
31
- "ف": 16,
32
- "ق": 36,
33
- "ك": 5,
34
- "ل": 22,
35
- "م": 11,
36
- "ن": 25,
37
- "ه": 24,
38
- "و": 39,
39
- "ى": 30,
40
- "ي": 41,
41
- "ٰ": 26,
42
- "ٱ": 13,
43
- "چ": 27,
44
- "ڤ": 4,
45
- "ک": 28,
46
- "ی": 1
47
  }
 
1
  {
2
+ "[PAD]": 42,
3
+ "[UNK]": 41,
4
+ "|": 1,
5
+ "ء": 21,
6
+ "آ": 28,
7
+ "أ": 29,
8
+ "ؤ": 11,
9
+ "إ": 7,
10
+ "ئ": 30,
11
+ "ا": 8,
12
+ "ب": 38,
13
+ "ة": 5,
14
+ "ت": 22,
15
+ "ث": 15,
16
+ "ج": 37,
17
+ "ح": 20,
18
+ "خ": 35,
19
+ "د": 32,
20
+ "ذ": 4,
21
+ "ر": 25,
22
+ "ز": 3,
23
+ "س": 0,
24
+ "ش": 17,
25
+ "ص": 34,
26
+ "ض": 14,
27
+ "ط": 9,
28
+ "ظ": 16,
29
+ "ع": 36,
30
+ "غ": 40,
31
+ "ف": 31,
32
+ "ق": 18,
33
+ "ك": 39,
34
+ "ل": 27,
35
+ "م": 2,
36
+ "ن": 13,
37
+ "ه": 6,
38
+ "و": 26,
39
+ "ى": 12,
40
+ "ي": 23,
41
+ "ٱ": 24,
42
+ "چ": 33,
43
+ "ڤ": 19,
44
+ "ۖ": 10
 
 
45
  }