Bakht Ullah commited on
Commit
6102137
1 Parent(s): ddac2be

Upload tokenizer

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