YassineBenlaria commited on
Commit
6a41e95
1 Parent(s): ecfd107

Upload tokenizer

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