hadiqa123 commited on
Commit
f9f5095
1 Parent(s): 4d08810

add tokenizer

Browse files
Files changed (2) hide show
  1. added_tokens.json +2 -2
  2. vocab.json +56 -65
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 67,
3
- "<s>": 66
4
  }
 
1
  {
2
+ "</s>": 58,
3
+ "<s>": 57
4
  }
vocab.json CHANGED
@@ -1,68 +1,59 @@
1
  {
2
- "'": 1,
3
- "[PAD]": 65,
4
- "[UNK]": 64,
5
  "|": 0,
6
- "،": 2,
7
- "ء": 3,
8
- "آ": 4,
9
- "أ": 5,
10
- "ؤ": 6,
11
- "ئ": 7,
12
- "ا": 8,
13
- "ب": 9,
14
- "ت": 10,
15
- "ث": 11,
16
- "ج": 12,
17
- "ح": 13,
18
- "خ": 14,
19
- "د": 15,
20
- "ذ": 16,
21
- "ر": 17,
22
- "ز": 18,
23
- "س": 19,
24
- "ش": 20,
25
- "ص": 21,
26
- "ض": 22,
27
- "ط": 23,
28
- "ظ": 24,
29
- "ع": 25,
30
- "غ": 26,
31
- "ف": 27,
32
- "ق": 28,
33
- "ل": 29,
34
- "م": 30,
35
- "ن": 31,
36
- "و": 32,
37
- "ى": 33,
38
- "ي": 34,
39
- "ً": 35,
40
- "َ": 36,
41
- "ُ": 37,
42
- "ِ": 38,
43
- "ّ": 39,
44
- "ٓ": 40,
45
- "ٔ": 41,
46
- "ٰ": 42,
47
- "ٹ": 43,
48
- "پ": 44,
49
- "چ": 45,
50
- "ڈ": 46,
51
- "ڑ": 47,
52
- "ژ": 48,
53
- "ک": 49,
54
- "گ": 50,
55
- "ں": 51,
56
- "ھ": 52,
57
- "ہ": 53,
58
- "ۂ": 54,
59
- "ۃ": 55,
60
- "ی": 56,
61
- "ے": 57,
62
- "ۓ": 58,
63
- "۔": 59,
64
- "’": 60,
65
- "…": 61,
66
- "ﷲ": 62,
67
- "ﷺ": 63
68
  }
 
1
  {
2
+ "[PAD]": 56,
3
+ "[UNK]": 55,
 
4
  "|": 0,
5
+ "،": 1,
6
+ "ء": 2,
7
+ "آ": 3,
8
+ "أ": 4,
9
+ "ؤ": 5,
10
+ "ئ": 6,
11
+ "ا": 7,
12
+ "ب": 8,
13
+ "ت": 9,
14
+ "ث": 10,
15
+ "ج": 11,
16
+ "ح": 12,
17
+ "خ": 13,
18
+ "د": 14,
19
+ "ذ": 15,
20
+ "ر": 16,
21
+ "ز": 17,
22
+ "س": 18,
23
+ "ش": 19,
24
+ "ص": 20,
25
+ "ض": 21,
26
+ "ط": 22,
27
+ "ظ": 23,
28
+ "ع": 24,
29
+ "غ": 25,
30
+ "ف": 26,
31
+ "ق": 27,
32
+ "ل": 28,
33
+ "م": 29,
34
+ "ن": 30,
35
+ "و": 31,
36
+ "ى": 32,
37
+ "ي": 33,
38
+ "ّ": 34,
39
+ "ٹ": 35,
40
+ "پ": 36,
41
+ "چ": 37,
42
+ "ڈ": 38,
43
+ "ڑ": 39,
44
+ "ژ": 40,
45
+ "ک": 41,
46
+ "گ": 42,
47
+ "ں": 43,
48
+ "ھ": 44,
49
+ "ہ": 45,
50
+ "ۂ": 46,
51
+ "ۃ": 47,
52
+ "ی": 48,
53
+ "ے": 49,
54
+ "ۓ": 50,
55
+ "۔": 51,
56
+ "": 52,
57
+ "": 53,
58
+ "": 54
 
 
 
 
 
 
 
 
59
  }