Hugging Face's logo

hts98
/
wav2vec2-xls-r-300m-paper

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
hts98/original_ver1.2
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
hts98 commited on
Commit
1ab1344
·
1 Parent(s): e9241c1

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +102 -102
vocab.json CHANGED
@@ -1,108 +1,108 @@
1
  {
2
- "0": 85,
3
- "1": 1,
4
- "2": 76,
5
- "3": 31,
6
- "4": 69,
7
- "5": 13,
8
- "6": 5,
9
- "7": 17,
10
- "8": 53,
11
- "9": 101,
12
  "[PAD]": 105,
13
  "[UNK]": 104,
14
- "a": 79,
15
- "b": 0,
16
- "c": 11,
17
- "d": 6,
18
- "e": 54,
19
- "f": 68,
20
- "g": 29,
21
- "h": 99,
22
- "i": 92,
23
- "j": 44,
24
- "k": 72,
25
- "l": 36,
26
- "m": 95,
27
- "n": 103,
28
- "o": 23,
29
- "p": 73,
30
- "q": 8,
31
- "r": 42,
32
- "s": 49,
33
- "t": 37,
34
- "u": 47,
35
- "v": 9,
36
- "w": 80,
37
- "x": 14,
38
- "y": 83,
39
- "z": 19,
40
- "|": 15,
41
- "à": 24,
42
- "á": 56,
43
- "â": 18,
44
- "ã": 34,
45
  "è": 26,
46
- "é": 70,
47
- "ê": 48,
48
- "ì": 38,
49
- "í": 94,
50
- "ò": 10,
51
- "ó": 21,
52
- "ô": 12,
53
- "õ": 51,
54
- "ù": 20,
55
- "ú": 61,
56
- "ý": 28,
57
- "ă": 97,
58
- "đ": 64,
59
- "ĩ": 71,
60
- "ũ": 58,
61
- "ơ": 90,
62
- "ư": 77,
63
- "ạ": 59,
64
- "ả": 98,
65
- "ấ": 27,
66
- "ầ": 67,
67
- "ẩ": 39,
68
- "ẫ": 55,
69
- "ậ": 35,
70
- "ắ": 4,
71
- "ằ": 52,
72
- "ẳ": 81,
73
- "ẵ": 57,
74
- "ặ": 93,
75
- "ẹ": 32,
76
  "ẻ": 78,
77
- "ẽ": 100,
78
- "ế": 3,
79
- "ề": 40,
80
- "ể": 25,
81
- "ễ": 63,
82
- "ệ": 87,
83
- "ỉ": 41,
84
- "ị": 102,
85
- "ọ": 84,
86
- "ỏ": 22,
87
- "ố": 66,
88
- "ồ": 88,
89
- "ổ": 33,
90
- "ỗ": 46,
91
- "ộ": 7,
92
- "ớ": 75,
93
- "ờ": 60,
94
- "ở": 74,
95
- "ỡ": 50,
96
- "ợ": 45,
97
- "ụ": 89,
98
- "ủ": 30,
99
- "ứ": 96,
100
- "ừ": 86,
101
- "ử": 65,
102
- "ữ": 2,
103
- "ự": 16,
104
- "ỳ": 43,
105
- "ỷ": 82,
106
- "ỹ": 62,
107
- "₫": 91
108
  }
 
1
  {
2
+ "0": 58,
3
+ "1": 82,
4
+ "2": 8,
5
+ "3": 101,
6
+ "4": 41,
7
+ "5": 17,
8
+ "6": 98,
9
+ "7": 102,
10
+ "8": 43,
11
+ "9": 80,
12
  "[PAD]": 105,
13
  "[UNK]": 104,
14
+ "a": 18,
15
+ "b": 56,
16
+ "c": 6,
17
+ "d": 53,
18
+ "e": 42,
19
+ "f": 5,
20
+ "g": 36,
21
+ "h": 0,
22
+ "i": 2,
23
+ "j": 55,
24
+ "k": 61,
25
+ "l": 22,
26
+ "m": 12,
27
+ "n": 76,
28
+ "o": 68,
29
+ "p": 7,
30
+ "q": 81,
31
+ "r": 91,
32
+ "s": 57,
33
+ "t": 84,
34
+ "u": 45,
35
+ "v": 28,
36
+ "w": 69,
37
+ "x": 25,
38
+ "y": 85,
39
+ "z": 20,
40
+ "|": 63,
41
+ "à": 52,
42
+ "á": 19,
43
+ "â": 13,
44
+ "ã": 62,
45
  "è": 26,
46
+ "é": 23,
47
+ "ê": 86,
48
+ "ì": 65,
49
+ "í": 79,
50
+ "ò": 49,
51
+ "ó": 100,
52
+ "ô": 4,
53
+ "õ": 77,
54
+ "ù": 90,
55
+ "ú": 33,
56
+ "ý": 50,
57
+ "ă": 93,
58
+ "đ": 72,
59
+ "ĩ": 96,
60
+ "ũ": 47,
61
+ "ơ": 66,
62
+ "ư": 99,
63
+ "ạ": 97,
64
+ "ả": 67,
65
+ "ấ": 16,
66
+ "ầ": 38,
67
+ "ẩ": 71,
68
+ "ẫ": 73,
69
+ "ậ": 83,
70
+ "ắ": 29,
71
+ "ằ": 30,
72
+ "ẳ": 3,
73
+ "ẵ": 10,
74
+ "ặ": 14,
75
+ "ẹ": 27,
76
  "ẻ": 78,
77
+ "ẽ": 46,
78
+ "ế": 32,
79
+ "ề": 24,
80
+ "ể": 34,
81
+ "ễ": 48,
82
+ "ệ": 39,
83
+ "ỉ": 89,
84
+ "ị": 94,
85
+ "ọ": 35,
86
+ "ỏ": 37,
87
+ "ố": 1,
88
+ "ồ": 59,
89
+ "ổ": 92,
90
+ "ỗ": 51,
91
+ "ộ": 64,
92
+ "ớ": 21,
93
+ "ờ": 95,
94
+ "ở": 88,
95
+ "ỡ": 74,
96
+ "ợ": 11,
97
+ "ụ": 44,
98
+ "ủ": 54,
99
+ "ứ": 87,
100
+ "ừ": 60,
101
+ "ử": 75,
102
+ "ữ": 70,
103
+ "ự": 103,
104
+ "ỳ": 9,
105
+ "ỷ": 15,
106
+ "ỹ": 40,
107
+ "₫": 31
108
  }