amansingh203 commited on
Commit
79f26e3
β€’
1 Parent(s): 438f425

Upload tokenizer

Browse files
Files changed (2) hide show
  1. tokenizer_config.json +0 -3
  2. vocab.json +35 -35
tokenizer_config.json CHANGED
@@ -1,12 +1,9 @@
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": "|"
 
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": "|"
vocab.json CHANGED
@@ -1,39 +1,39 @@
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
- "Γ­": 29,
35
- "β€”": 16,
36
- "’": 4,
37
- "β€œ": 23,
38
- "”": 5
39
  }
 
1
  {
2
+ "'": 22,
3
+ "(": 32,
4
+ ")": 0,
5
  "[PAD]": 36,
6
  "[UNK]": 35,
7
+ "a": 13,
8
+ "b": 23,
9
+ "c": 4,
10
+ "d": 18,
11
+ "e": 8,
12
+ "f": 6,
13
+ "g": 2,
14
+ "h": 9,
15
+ "i": 10,
16
+ "j": 21,
17
+ "k": 14,
18
+ "l": 17,
19
+ "m": 1,
20
+ "n": 11,
21
+ "o": 25,
22
+ "p": 7,
23
+ "q": 15,
24
+ "r": 30,
25
+ "s": 12,
26
+ "t": 16,
27
+ "u": 27,
28
+ "v": 28,
29
+ "w": 33,
30
+ "x": 20,
31
+ "y": 31,
32
+ "z": 5,
33
+ "|": 34,
34
+ "Γ­": 19,
35
+ "β€”": 26,
36
+ "’": 3,
37
+ "β€œ": 29,
38
+ "”": 24
39
  }