ad019el commited on
Commit
b190b8c
1 Parent(s): 3ed8865

Upload tokenizer

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