YassineBenlaria commited on
Commit
2e3aae9
1 Parent(s): d68865a

Upload tokenizer

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