ad019el commited on
Commit
a478c86
1 Parent(s): b615361

add tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +61 -61
vocab.json CHANGED
@@ -1,66 +1,66 @@
1
  {
2
- "/": 28,
3
- "0": 12,
4
- "1": 26,
5
- "2": 39,
6
- "7": 6,
7
- "9": 34,
8
  "[PAD]": 63,
9
  "[UNK]": 62,
10
- "_": 52,
11
- "a": 36,
12
- "c": 20,
13
- "e": 8,
14
- "f": 56,
15
- "j": 45,
16
- "l": 42,
17
- "m": 48,
18
- "o": 23,
19
- "q": 49,
20
- "r": 1,
21
- "t": 47,
22
- "u": 33,
23
- "w": 54,
24
- "|": 10,
25
- "ء": 38,
26
- "آ": 29,
27
  "أ": 22,
28
- "ؤ": 44,
29
- "إ": 24,
30
- "ئ": 9,
31
- "ا": 7,
32
- "ب": 30,
33
- "ة": 27,
34
- "ت": 37,
35
- "ث": 50,
36
- "ج": 35,
37
- "ح": 46,
38
- "خ": 19,
39
- "د": 5,
40
- "ذ": 14,
41
- "ر": 41,
42
- "ز": 15,
43
- "س": 13,
44
- "ش": 0,
45
- "ص": 55,
46
- "ض": 16,
47
- "ط": 11,
48
- "ظ": 53,
49
- "ع": 58,
50
- "غ": 59,
51
- "ف": 21,
52
- "ق": 25,
53
- "ك": 60,
54
- "ل": 4,
55
- "م": 51,
56
- "ن": 32,
57
- "ه": 31,
58
- "و": 61,
59
- "ى": 3,
60
- "ي": 43,
61
- "ٱ": 18,
62
- "چ": 40,
63
- "ڤ": 2,
64
- "ک": 57,
65
- "ی": 17
66
  }
 
1
  {
2
+ "/": 31,
3
+ "0": 3,
4
+ "1": 39,
5
+ "2": 55,
6
+ "7": 15,
7
+ "9": 21,
8
  "[PAD]": 63,
9
  "[UNK]": 62,
10
+ "_": 47,
11
+ "a": 29,
12
+ "c": 34,
13
+ "e": 53,
14
+ "f": 11,
15
+ "j": 6,
16
+ "l": 58,
17
+ "m": 50,
18
+ "o": 5,
19
+ "q": 46,
20
+ "r": 45,
21
+ "t": 17,
22
+ "u": 23,
23
+ "w": 7,
24
+ "|": 25,
25
+ "ء": 8,
26
+ "آ": 44,
27
  "أ": 22,
28
+ "ؤ": 40,
29
+ "إ": 36,
30
+ "ئ": 41,
31
+ "ا": 24,
32
+ "ب": 16,
33
+ "ة": 12,
34
+ "ت": 30,
35
+ "ث": 2,
36
+ "ج": 56,
37
+ "ح": 37,
38
+ "خ": 43,
39
+ "د": 26,
40
+ "ذ": 33,
41
+ "ر": 32,
42
+ "ز": 35,
43
+ "س": 42,
44
+ "ش": 59,
45
+ "ص": 54,
46
+ "ض": 4,
47
+ "ط": 13,
48
+ "ظ": 49,
49
+ "ع": 19,
50
+ "غ": 61,
51
+ "ف": 27,
52
+ "ق": 20,
53
+ "ك": 52,
54
+ "ل": 9,
55
+ "م": 14,
56
+ "ن": 38,
57
+ "ه": 0,
58
+ "و": 57,
59
+ "ى": 48,
60
+ "ي": 28,
61
+ "ٱ": 1,
62
+ "چ": 60,
63
+ "ڤ": 18,
64
+ "ک": 10,
65
+ "ی": 51
66
  }