Hugging Face's logo

yip-i
/
colab-demo

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
Generated from Trainer
Model card Files
xet
 Metrics Community
1
yip-i commited on
Commit
783031b
·
1 Parent(s): 26655df

Upload tokenizer

Browse files
Files changed (3) hide show
  1. special_tokens_map.json +2 -0
  2. tokenizer_config.json +2 -2
  3. vocab.json +30 -30
special_tokens_map.json CHANGED
@@ -1,4 +1,6 @@
1
  {
 
 
2
  "pad_token": "[PAD]",
3
  "unk_token": "[UNK]"
4
  }
 
1
  {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
  "pad_token": "[PAD]",
5
  "unk_token": "[UNK]"
6
  }
tokenizer_config.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
- "bos_token": null,
3
  "do_lower_case": false,
4
- "eos_token": null,
5
  "pad_token": "[PAD]",
6
  "replace_word_delimiter_char": " ",
7
  "tokenizer_class": "Wav2Vec2CTCTokenizer",
 
1
  {
2
+ "bos_token": "<s>",
3
  "do_lower_case": false,
4
+ "eos_token": "</s>",
5
  "pad_token": "[PAD]",
6
  "replace_word_delimiter_char": " ",
7
  "tokenizer_class": "Wav2Vec2CTCTokenizer",
vocab.json CHANGED
@@ -1,34 +1,34 @@
1
  {
2
- "'": 20,
3
- "1": 26,
4
- "3": 21,
5
  "[PAD]": 31,
6
  "[UNK]": 30,
7
- "a": 14,
8
- "b": 10,
9
- "c": 6,
10
- "d": 2,
11
- "e": 12,
12
- "f": 18,
13
- "g": 7,
14
- "h": 29,
15
- "i": 22,
16
- "j": 0,
17
- "k": 11,
18
- "l": 4,
19
- "m": 8,
20
- "n": 15,
21
- "o": 17,
22
- "p": 24,
23
- "q": 19,
24
- "r": 28,
25
- "s": 23,
26
- "t": 16,
27
- "u": 13,
28
- "v": 27,
29
- "w": 1,
30
- "x": 9,
31
- "y": 25,
32
- "z": 3,
33
- "|": 5
34
  }
 
1
  {
2
+ "'": 14,
3
+ "1": 24,
4
+ "3": 15,
5
  "[PAD]": 31,
6
  "[UNK]": 30,
7
+ "a": 23,
8
+ "b": 16,
9
+ "c": 11,
10
+ "d": 12,
11
+ "e": 9,
12
+ "f": 20,
13
+ "g": 17,
14
+ "h": 2,
15
+ "i": 5,
16
+ "j": 4,
17
+ "k": 3,
18
+ "l": 13,
19
+ "m": 26,
20
+ "n": 21,
21
+ "o": 22,
22
+ "p": 0,
23
+ "q": 18,
24
+ "r": 6,
25
+ "s": 7,
26
+ "t": 19,
27
+ "u": 25,
28
+ "v": 1,
29
+ "w": 10,
30
+ "x": 27,
31
+ "y": 8,
32
+ "z": 28,
33
+ "|": 29
34
  }