Hugging Face's logo

shishirml
/
wav2vec2-large-mms-1b-Nepali-colab-Shishir

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
Inference Endpoints
Model card Files Metrics Community
1
shishirml commited on
Commit
d37197f
•
1 Parent(s): d8f33c2

Upload tokenizer

Browse files
Files changed (2) hide show
  1. added_tokens.json +2 -2
  2. vocab.json +76 -85
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 87,
3
- "<s>": 86
4
  }
 
1
  {
2
+ "</s>": 78,
3
+ "<s>": 77
4
  }
vocab.json CHANGED
@@ -1,90 +1,81 @@
1
  {
2
  "npi": {
3
- "[PAD]": 85,
4
- "[UNK]": 84,
5
- "\\": 1,
6
  "|": 0,
7
- "ँ": 2,
8
- "ं": 3,
9
- "ः": 4,
10
- "अ": 5,
11
- "आ": 6,
12
- "इ": 7,
13
- "ई": 8,
14
- "उ": 9,
15
- "ऊ": 10,
16
- "ऋ": 11,
17
- "ए": 12,
18
- "ऐ": 13,
19
- "ओ": 14,
20
- "औ": 15,
21
- "क": 16,
22
- "ख": 17,
23
- "ग": 18,
24
- "घ": 19,
25
- "ङ": 20,
26
- "च": 21,
27
- "छ": 22,
28
- "ज": 23,
29
- "झ": 24,
30
- "ञ": 25,
31
- "ट": 26,
32
- "ठ": 27,
33
- "ड": 28,
34
- "ढ": 29,
35
- "ण": 30,
36
- "त": 31,
37
- "थ": 32,
38
- "द": 33,
39
- "ध": 34,
40
- "न": 35,
41
- "प": 36,
42
- "फ": 37,
43
- "ब": 38,
44
- "भ": 39,
45
- "म": 40,
46
- "य": 41,
47
- "र": 42,
48
- "ऱ": 43,
49
- "ल": 44,
50
- "व": 45,
51
- "श": 46,
52
- "ष": 47,
53
- "स": 48,
54
- "ह": 49,
55
- "़": 50,
56
- "ा": 51,
57
- "ि": 52,
58
- "ी": 53,
59
- "ु": 54,
60
- "ू": 55,
61
- "ृ": 56,
62
- "े": 57,
63
- "ै": 58,
64
- "ॉ": 59,
65
- "ॊ": 60,
66
- "ो": 61,
67
- "ौ": 62,
68
- "्": 63,
69
- "॑": 64,
70
- "ॠ": 65,
71
- "।": 66,
72
- "०": 67,
73
- "१": 68,
74
- "२": 69,
75
- "३": 70,
76
- "४": 71,
77
- "५": 72,
78
- "६": 73,
79
- "७": 74,
80
- "८": 75,
81
- "९": 76,
82
- "॰": 77,
83
- "​": 78,
84
- "‌": 79,
85
- "‍": 80,
86
- "‎": 81,
87
- "‏": 82,
88
- " ": 83
89
  }
90
  }
 
1
  {
2
  "npi": {
3
+ "[PAD]": 76,
4
+ "[UNK]": 75,
 
5
  "|": 0,
6
+ "ँ": 1,
7
+ "ं": 2,
8
+ "ः": 3,
9
+ "अ": 4,
10
+ "आ": 5,
11
+ "इ": 6,
12
+ "ई": 7,
13
+ "उ": 8,
14
+ "ऊ": 9,
15
+ "ऋ": 10,
16
+ "ए": 11,
17
+ "ऐ": 12,
18
+ "ओ": 13,
19
+ "औ": 14,
20
+ "क": 15,
21
+ "ख": 16,
22
+ "ग": 17,
23
+ "घ": 18,
24
+ "ङ": 19,
25
+ "च": 20,
26
+ "छ": 21,
27
+ "ज": 22,
28
+ "झ": 23,
29
+ "ञ": 24,
30
+ "ट": 25,
31
+ "ठ": 26,
32
+ "ड": 27,
33
+ "ढ": 28,
34
+ "ण": 29,
35
+ "त": 30,
36
+ "थ": 31,
37
+ "द": 32,
38
+ "ध": 33,
39
+ "न": 34,
40
+ "प": 35,
41
+ "फ": 36,
42
+ "ब": 37,
43
+ "भ": 38,
44
+ "म": 39,
45
+ "य": 40,
46
+ "र": 41,
47
+ "ल": 42,
48
+ "व": 43,
49
+ "श": 44,
50
+ "ष": 45,
51
+ "स": 46,
52
+ "ह": 47,
53
+ "़": 48,
54
+ "ा": 49,
55
+ "ि": 50,
56
+ "ी": 51,
57
+ "ु": 52,
58
+ "ू": 53,
59
+ "ृ": 54,
60
+ "े": 55,
61
+ "ै": 56,
62
+ "ॉ": 57,
63
+ "ो": 58,
64
+ "ौ": 59,
65
+ "्": 60,
66
+ "ॠ": 61,
67
+ "।": 62,
68
+ "०": 63,
69
+ "१": 64,
70
+ "२": 65,
71
+ "३": 66,
72
+ "४": 67,
73
+ "५": 68,
74
+ "६": 69,
75
+ "७": 70,
76
+ "८": 71,
77
+ "९": 72,
78
+ "‌": 73,
79
+ "‍": 74
 
 
 
 
 
 
 
 
80
  }
81
  }