ad019el commited on
Commit
815b491
1 Parent(s): 524d392

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