Hugging Face's logo

gweltou
/
wav2vec2-large-xlsr-53-br

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
gweltou commited on
Commit
9e26652
•
1 Parent(s): b366b6f

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +48 -40
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 41,
3
- "<s>": 40
4
  }
 
1
  {
2
+ "</s>": 49,
3
+ "<s>": 48
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "38": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "39": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "40": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "41": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "46": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "47": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "48": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "49": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,42 +1,50 @@
1
  {
2
- "'": 17,
3
- "[PAD]": 39,
4
- "[UNK]": 38,
5
- "a": 12,
6
- "b": 14,
7
- "c": 7,
8
- "d": 15,
9
- "e": 0,
10
- "f": 21,
11
- "g": 2,
12
- "h": 9,
13
- "i": 29,
14
- "j": 37,
15
- "k": 11,
16
- "l": 30,
17
- "m": 13,
18
- "n": 32,
19
- "o": 24,
20
- "p": 25,
21
- "q": 27,
22
- "r": 1,
23
- "s": 23,
24
- "t": 22,
25
- "u": 20,
26
- "v": 28,
27
- "w": 5,
28
- "x": 4,
29
- "y": 3,
30
- "z": 31,
31
- "|": 8,
32
- "à": 34,
33
- "â": 26,
34
- "è": 33,
35
- "é": 35,
36
- "ê": 16,
37
- "ñ": 6,
38
- "ô": 10,
39
- "ù": 36,
40
- "û": 18,
41
- "ü": 19
 
 
 
 
 
 
 
 
42
  }
 
1
  {
2
+ "'": 38,
3
+ "3": 24,
4
+ "[PAD]": 47,
5
+ "[UNK]": 46,
6
+ "a": 43,
7
+ "b": 11,
8
+ "c": 4,
9
+ "d": 13,
10
+ "e": 34,
11
+ "f": 28,
12
+ "g": 33,
13
+ "h": 20,
14
+ "i": 7,
15
+ "j": 14,
16
+ "k": 22,
17
+ "l": 19,
18
+ "m": 35,
19
+ "n": 30,
20
+ "o": 41,
21
+ "p": 9,
22
+ "q": 17,
23
+ "r": 26,
24
+ "s": 21,
25
+ "t": 31,
26
+ "u": 5,
27
+ "v": 27,
28
+ "w": 3,
29
+ "x": 37,
30
+ "y": 6,
31
+ "z": 40,
32
+ "|": 1,
33
+ "à": 42,
34
+ "á": 2,
35
+ "â": 29,
36
+ "ã": 8,
37
+ "ç": 25,
38
+ "è": 16,
39
+ "é": 23,
40
+ "ê": 36,
41
+ "ë": 18,
42
+ "ñ": 0,
43
+ "ô": 44,
44
+ "ö": 10,
45
+ "ø": 39,
46
+ "ù": 32,
47
+ "û": 45,
48
+ "ü": 15,
49
+ "ı": 12
50
  }