Hugging Face's logo

amansingh203
/
wav2vec2

Model card Files Community
amansingh203 commited on
Commit
438f425
•
1 Parent(s): ec72a27

Upload tokenizer

Browse files
Files changed (2) hide show
  1. tokenizer_config.json +3 -0
  2. vocab.json +34 -34
tokenizer_config.json CHANGED
@@ -1,9 +1,12 @@
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",
8
  "unk_token": "[UNK]",
9
  "word_delimiter_token": "|"
 
1
  {
2
  "bos_token": "<s>",
3
+ "clean_up_tokenization_spaces": true,
4
  "do_lower_case": false,
5
  "eos_token": "</s>",
6
+ "model_max_length": 1000000000000000019884624838656,
7
  "pad_token": "[PAD]",
8
  "replace_word_delimiter_char": " ",
9
+ "target_lang": null,
10
  "tokenizer_class": "Wav2Vec2CTCTokenizer",
11
  "unk_token": "[UNK]",
12
  "word_delimiter_token": "|"
vocab.json CHANGED
@@ -1,39 +1,39 @@
1
  {
2
- "'": 6,
3
- "(": 7,
4
- ")": 26,
5
  "[PAD]": 36,
6
  "[UNK]": 35,
7
- "a": 23,
8
- "b": 32,
9
- "c": 22,
10
- "d": 16,
11
- "e": 0,
12
- "f": 10,
13
- "g": 4,
14
- "h": 30,
15
- "i": 11,
16
- "j": 9,
17
- "k": 2,
18
- "l": 31,
19
- "m": 24,
20
  "n": 14,
21
- "o": 34,
22
- "p": 19,
23
- "q": 15,
24
- "r": 3,
25
- "s": 20,
26
- "t": 13,
27
- "u": 5,
28
- "v": 33,
29
- "w": 12,
30
- "x": 27,
31
- "y": 21,
32
- "z": 18,
33
- "|": 17,
34
- "í": 28,
35
- "—": 1,
36
- "’": 29,
37
- "“": 8,
38
- "”": 25
39
  }
 
1
  {
2
+ "'": 0,
3
+ "(": 8,
4
+ ")": 1,
5
  "[PAD]": 36,
6
  "[UNK]": 35,
7
+ "a": 6,
8
+ "b": 31,
9
+ "c": 7,
10
+ "d": 10,
11
+ "e": 15,
12
+ "f": 21,
13
+ "g": 13,
14
+ "h": 25,
15
+ "i": 19,
16
+ "j": 30,
17
+ "k": 11,
18
+ "l": 22,
19
+ "m": 28,
20
  "n": 14,
21
+ "o": 12,
22
+ "p": 3,
23
+ "q": 33,
24
+ "r": 24,
25
+ "s": 27,
26
+ "t": 32,
27
+ "u": 18,
28
+ "v": 26,
29
+ "w": 2,
30
+ "x": 17,
31
+ "y": 34,
32
+ "z": 20,
33
+ "|": 9,
34
+ "í": 29,
35
+ "—": 16,
36
+ "’": 4,
37
+ "“": 23,
38
+ "”": 5
39
  }