Hugging Face's logo

Sajjo
/
w2v-bert-2.0-bangala-gpu-CV16.0

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
wav2vec2-bert
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
Sajjo commited on
Commit
16cd6e6
•
1 Parent(s): f287a05

Upload processor

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +70 -69
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 73,
3
- "<s>": 72
4
  }
 
1
  {
2
+ "</s>": 74,
3
+ "<s>": 73
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "70": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "71": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "72": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "73": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "71": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "72": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "73": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "74": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,74 +1,75 @@
1
  {
2
  "/": 1,
3
- "[PAD]": 71,
4
- "[UNK]": 70,
5
  "|": 0,
6
  "।": 2,
7
- "ঁ": 3,
8
- "ং": 4,
9
- "ঃ": 5,
10
- "অ": 6,
11
- "আ": 7,
12
- "ই": 8,
13
- "ঈ": 9,
14
- "উ": 10,
15
- "ঊ": 11,
16
- "ঋ": 12,
17
- "এ": 13,
18
- "ঐ": 14,
19
- "ও": 15,
20
- "ঔ": 16,
21
- "ক": 17,
22
- "খ": 18,
23
- "গ": 19,
24
- "ঘ": 20,
25
- "ঙ": 21,
26
- "চ": 22,
27
- "ছ": 23,
28
- "জ": 24,
29
- "ঝ": 25,
30
- "ঞ": 26,
31
- "ট": 27,
32
- "ঠ": 28,
33
- "ড": 29,
34
- "ঢ": 30,
35
- "ণ": 31,
36
- "ত": 32,
37
- "থ": 33,
38
- "দ": 34,
39
- "ধ": 35,
40
- "ন": 36,
41
- "প": 37,
42
- "ফ": 38,
43
- "ব": 39,
44
- "ভ": 40,
45
- "ম": 41,
46
- "য": 42,
47
- "র": 43,
48
- "ল": 44,
49
- "শ": 45,
50
- "ষ": 46,
51
- "স": 47,
52
- "হ": 48,
53
- "়": 49,
54
- "া": 50,
55
- "ি": 51,
56
- "ী": 52,
57
- "ু": 53,
58
- "ূ": 54,
59
- "ৃ": 55,
60
- "ে": 56,
61
- "ৈ": 57,
62
- "ো": 58,
63
- "ৌ": 59,
64
- "্": 60,
65
- "ৎ": 61,
66
- "ড়": 62,
67
- "ঢ়": 63,
68
- "য়": 64,
69
- "ৰ": 65,
70
- "–": 66,
71
- "—": 67,
72
- "’": 68,
73
- "‚": 69
 
74
  }
 
1
  {
2
  "/": 1,
3
+ "[PAD]": 72,
4
+ "[UNK]": 71,
5
  "|": 0,
6
  "।": 2,
7
+ "॥": 3,
8
+ "ঁ": 4,
9
+ "ং": 5,
10
+ "ঃ": 6,
11
+ "অ": 7,
12
+ "আ": 8,
13
+ "ই": 9,
14
+ "ঈ": 10,
15
+ "উ": 11,
16
+ "ঊ": 12,
17
+ "ঋ": 13,
18
+ "এ": 14,
19
+ "ঐ": 15,
20
+ "ও": 16,
21
+ "ঔ": 17,
22
+ "ক": 18,
23
+ "খ": 19,
24
+ "গ": 20,
25
+ "ঘ": 21,
26
+ "ঙ": 22,
27
+ "চ": 23,
28
+ "ছ": 24,
29
+ "জ": 25,
30
+ "ঝ": 26,
31
+ "ঞ": 27,
32
+ "ট": 28,
33
+ "ঠ": 29,
34
+ "ড": 30,
35
+ "ঢ": 31,
36
+ "ণ": 32,
37
+ "ত": 33,
38
+ "থ": 34,
39
+ "দ": 35,
40
+ "ধ": 36,
41
+ "ন": 37,
42
+ "প": 38,
43
+ "ফ": 39,
44
+ "ব": 40,
45
+ "ভ": 41,
46
+ "ম": 42,
47
+ "য": 43,
48
+ "র": 44,
49
+ "ল": 45,
50
+ "শ": 46,
51
+ "ষ": 47,
52
+ "স": 48,
53
+ "হ": 49,
54
+ "়": 50,
55
+ "া": 51,
56
+ "ি": 52,
57
+ "ী": 53,
58
+ "ু": 54,
59
+ "ূ": 55,
60
+ "ৃ": 56,
61
+ "ে": 57,
62
+ "ৈ": 58,
63
+ "ো": 59,
64
+ "ৌ": 60,
65
+ "্": 61,
66
+ "ৎ": 62,
67
+ "ড়": 63,
68
+ "ঢ়": 64,
69
+ "য়": 65,
70
+ "ৰ": 66,
71
+ "–": 67,
72
+ "—": 68,
73
+ "’": 69,
74
+ "‚": 70
75
  }