ad019el commited on
Commit
bcdc7dd
1 Parent(s): a4b364d

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +42 -43
vocab.json CHANGED
@@ -1,46 +1,45 @@
1
  {
2
- "[PAD]": 43,
3
- "[UNK]": 42,
4
- "\xa0": 41
5
- "|": 8,
6
- "ء": 23,
7
- "آ": 34,
8
- "أ": 38,
9
- "ؤ": 10,
10
- "إ": 1,
11
  "ئ": 9,
12
- "ا": 4,
13
- "ب": 32,
14
- "ة": 20,
15
- "ت": 21,
16
- "ث": 19,
17
- "ج": 13,
18
- "ح": 37,
19
- "خ": 17,
20
- "د": 26,
21
- "ذ": 27,
22
- "ر": 2,
23
- "ز": 29,
24
- "س": 28,
25
- "ش": 0,
26
- "ص": 24,
27
- "ض": 33,
28
- "ط": 18,
29
- "ظ": 25,
30
- "ع": 7,
31
- "غ": 22,
32
- "ف": 15,
33
- "ق": 3,
34
- "ك": 6,
35
- "ل": 30,
36
- "م": 36,
37
- "ن": 40,
38
- "ه": 11,
39
- "و": 35,
40
- "ى": 31,
41
- "ي": 12,
42
- "ٱ": 39,
43
- "ڤ": 14,
44
- "ک": 5,
45
- "ی": 16
46
  }
 
1
  {
2
+ "[PAD]": 42,
3
+ "[UNK]": 41,
4
+ "|": 14,
5
+ "ء": 16,
6
+ "آ": 19,
7
+ "أ": 34,
8
+ "ؤ": 25,
9
+ "إ": 31,
 
10
  "ئ": 9,
11
+ "ا": 10,
12
+ "ب": 1,
13
+ "ة": 36,
14
+ "ت": 4,
15
+ "ث": 20,
16
+ "ج": 11,
17
+ "ح": 39,
18
+ "خ": 27,
19
+ "د": 8,
20
+ "ذ": 28,
21
+ "ر": 18,
22
+ "ز": 7,
23
+ "س": 22,
24
+ "ش": 26,
25
+ "ص": 40,
26
+ "ض": 29,
27
+ "ط": 6,
28
+ "ظ": 23,
29
+ "ع": 3,
30
+ "غ": 30,
31
+ "ف": 21,
32
+ "ق": 35,
33
+ "ك": 2,
34
+ "ل": 5,
35
+ "م": 0,
36
+ "ن": 32,
37
+ "ه": 15,
38
+ "و": 38,
39
+ "ى": 12,
40
+ "ي": 24,
41
+ "ٱ": 33,
42
+ "ڤ": 13,
43
+ "ک": 37,
44
+ "ی": 17
45
  }