nrshoudi commited on
Commit
bbc7d45
1 Parent(s): 089f809

Upload tokenizer

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