YassineBenlaria commited on
Commit
52b5d4c
1 Parent(s): 3015079

Upload tokenizer

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