Jawaria commited on
Commit
3379bcf
·
1 Parent(s): ea6a870

Upload tokenizer

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