ad019el commited on
Commit
2cbfb85
1 Parent(s): 5dba75f

Upload tokenizer

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