Hugging Face's logo

nbaden
/
wav2vec2-large-xlsr-53-demo-colab

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
nbaden commited on
Commit
c807540
•
1 Parent(s): 82743c5

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +4 -4
  2. tokenizer_config.json +4 -4
  3. vocab.json +45 -47
added_tokens.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
- "</s>": 49,
3
- "<s>": 48,
4
- "[PAD]": 47,
5
- "[UNK]": 46
6
  }
 
1
  {
2
+ "</s>": 47,
3
+ "<s>": 46,
4
+ "[PAD]": 45,
5
+ "[UNK]": 44
6
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "46": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "47": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "48": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "49": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "44": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "45": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "46": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "47": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,50 +1,48 @@
1
  {
2
- "[PAD]": 47,
3
- "[UNK]": 46,
4
- "c": 34,
5
- "i": 27,
6
- "o": 3,
7
- "y": 42,
8
- "|": 35,
9
- "«": 29,
10
- "»": 45,
11
- "а": 17,
12
- "б": 12,
13
- "в": 14,
14
- "г": 21,
15
- "д": 30,
16
- "е": 7,
17
- "ж": 20,
18
- "з": 16,
19
- "и": 9,
20
- "й": 41,
21
- "к": 11,
22
- "л": 25,
23
- "м": 13,
24
- "н": 26,
25
- "о": 15,
26
- "п": 37,
27
- "р": 5,
28
- "с": 38,
29
- "т": 33,
30
- "у": 2,
31
- "ф": 22,
32
- "х": 43,
33
- "ц": 6,
34
- "ч": 32,
35
- "ш": 0,
36
- "щ": 24,
37
- "ъ": 31,
38
  "ы": 39,
39
- "ь": 23,
40
- "э": 19,
41
- "ю": 28,
42
- "я": 1,
43
- "і": 4,
44
- "ј": 8,
45
- "ҕ": 10,
46
- "ҥ": 40,
47
- "ү": 18,
48
- "һ": 44,
49
- "ө": 36
50
  }
 
1
  {
2
+ "[PAD]": 45,
3
+ "[UNK]": 44,
4
+ "c": 20,
5
+ "i": 15,
6
+ "o": 30,
7
+ "y": 35,
8
+ "|": 36,
9
+ "«": 37,
10
+ "»": 31,
11
+ "а": 18,
12
+ "б": 22,
13
+ "в": 23,
14
+ "г": 42,
15
+ "д": 41,
16
+ "е": 17,
17
+ "ж": 2,
18
+ "з": 32,
19
+ "и": 0,
20
+ "й": 34,
21
+ "к": 3,
22
+ "л": 38,
23
+ "м": 14,
24
+ "н": 28,
25
+ "о": 33,
26
+ "п": 12,
27
+ "р": 11,
28
+ "с": 1,
29
+ "т": 24,
30
+ "у": 7,
31
+ "ф": 40,
32
+ "х": 16,
33
+ "ц": 9,
34
+ "ч": 21,
35
+ "ш": 6,
36
+ "щ": 19,
 
37
  "ы": 39,
38
+ "ь": 27,
39
+ "э": 8,
40
+ "ю": 5,
41
+ "я": 4,
42
+ "і": 26,
43
+ "ҕ": 43,
44
+ "ҥ": 25,
45
+ "ү": 13,
46
+ "һ": 10,
47
+ "ө": 29
 
48
  }