ad019el commited on
Commit
6a9fee1
1 Parent(s): 306e457

Upload tokenizer

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