ad019el commited on
Commit
1092c0e
1 Parent(s): 4c0dcf7

Upload tokenizer

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