Hugging Face's logo

hadiqa123
/
xlsr_ur_training

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
2
hadiqa123 commited on
Commit
e9b7ed9
•
1 Parent(s): ab17f4c

add tokenizer

Browse files
Files changed (2) hide show
  1. added_tokens.json +2 -2
  2. vocab.json +61 -53
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 59,
3
- "<s>": 58
4
  }
 
1
  {
2
+ "</s>": 67,
3
+ "<s>": 66
4
  }
vocab.json CHANGED
@@ -1,60 +1,68 @@
1
  {
2
- "[PAD]": 57,
3
- "[UNK]": 56,
4
  "|": 0,
5
  "،": 1,
6
  "؟": 2,
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
- "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
- "ُ": 35,
40
- "ِ": 36,
41
- "ّ": 37,
42
- "ٓ": 38,
43
- "ٔ": 39,
44
- "ٰ": 40,
45
- "ٹ": 41,
46
- "پ": 42,
47
- "چ": 43,
48
- "ڈ": 44,
49
- "ڑ": 45,
50
- "ژ": 46,
51
- "ک": 47,
52
- "گ": 48,
53
- "ں": 49,
54
- "ھ": 50,
55
- "ہ": 51,
56
- "ۂ": 52,
57
- "ی": 53,
58
- "ے": 54,
59
- "۔": 55
 
 
 
 
 
 
 
 
60
  }
 
1
  {
2
+ "[PAD]": 65,
3
+ "[UNK]": 64,
4
  "|": 0,
5
  "،": 1,
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
  }