ad019el commited on
Commit
532c863
1 Parent(s): 506a1cf

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