ad019el commited on
Commit
f185b10
1 Parent(s): 316e0e3

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