adid commited on
Commit
98cd8a0
1 Parent(s): d6ad665

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +45 -45
vocab.json CHANGED
@@ -1,49 +1,49 @@
1
  {
2
  "[PAD]": 46,
3
  "[UNK]": 45,
4
- "|": 35,
5
- "ء": 37,
6
- "آ": 21,
7
- "أ": 32,
8
- "ؤ": 28,
9
- "إ": 42,
10
- "ئ": 31,
11
- "ا": 22,
12
- "ب": 18,
13
- "ة": 38,
14
- "ت": 39,
15
- "ث": 0,
16
- "ج": 6,
17
- "ح": 33,
18
- "خ": 12,
19
- "د": 15,
20
- "ذ": 40,
21
- "ر": 24,
22
- "ز": 3,
23
- "س": 4,
24
- "ش": 10,
25
- "ص": 8,
26
- "ض": 9,
27
- "ط": 26,
28
- "ظ": 1,
29
- "ع": 11,
30
- "غ": 23,
31
- "ف": 7,
32
- "ق": 30,
33
- "ك": 5,
34
- "ل": 41,
35
- "م": 43,
36
- "ن": 27,
37
- "ه": 29,
38
- "و": 20,
39
- "ى": 16,
40
- "ي": 19,
41
- "ً": 36,
42
- "ٌ": 17,
43
- "ٍ": 13,
44
- "َ": 34,
45
- "ُ": 44,
46
- "ِ": 14,
47
- "ّ": 2,
48
- "ْ": 25
49
  }
 
1
  {
2
  "[PAD]": 46,
3
  "[UNK]": 45,
4
+ "|": 21,
5
+ "ء": 10,
6
+ "آ": 3,
7
+ "أ": 17,
8
+ "ؤ": 23,
9
+ "إ": 0,
10
+ "ئ": 37,
11
+ "ا": 20,
12
+ "ب": 1,
13
+ "ة": 40,
14
+ "ت": 9,
15
+ "ث": 34,
16
+ "ج": 5,
17
+ "ح": 32,
18
+ "خ": 29,
19
+ "د": 2,
20
+ "ذ": 27,
21
+ "ر": 43,
22
+ "ز": 12,
23
+ "س": 7,
24
+ "ش": 36,
25
+ "ص": 31,
26
+ "ض": 11,
27
+ "ط": 15,
28
+ "ظ": 6,
29
+ "ع": 35,
30
+ "غ": 38,
31
+ "ف": 41,
32
+ "ق": 19,
33
+ "ك": 26,
34
+ "ل": 30,
35
+ "م": 22,
36
+ "ن": 14,
37
+ "ه": 8,
38
+ "و": 25,
39
+ "ى": 42,
40
+ "ي": 28,
41
+ "ً": 13,
42
+ "ٌ": 18,
43
+ "ٍ": 39,
44
+ "َ": 44,
45
+ "ُ": 33,
46
+ "ِ": 16,
47
+ "ّ": 24,
48
+ "ْ": 4
49
  }