khaoulaoub commited on
Commit
e2f38b4
1 Parent(s): 88b2c43

Upload tokenizer

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