Cantaosu commited on
Commit
1051ca2
·
verified ·
1 Parent(s): 60601c3

Upload tokenizer

Browse files
Files changed (2) hide show
  1. tokenizer_config.json +0 -1
  2. vocab.json +27 -27
tokenizer_config.json CHANGED
@@ -39,7 +39,6 @@
39
  "eos_token": "</s>",
40
  "model_max_length": 1000000000000000019884624838656,
41
  "pad_token": "[PAD]",
42
- "processor_class": "Wav2Vec2Processor",
43
  "replace_word_delimiter_char": " ",
44
  "target_lang": null,
45
  "tokenizer_class": "Wav2Vec2CTCTokenizer",
 
39
  "eos_token": "</s>",
40
  "model_max_length": 1000000000000000019884624838656,
41
  "pad_token": "[PAD]",
 
42
  "replace_word_delimiter_char": " ",
43
  "target_lang": null,
44
  "tokenizer_class": "Wav2Vec2CTCTokenizer",
vocab.json CHANGED
@@ -1,32 +1,32 @@
1
  {
2
- "'": 9,
3
  "[PAD]": 29,
4
  "[UNK]": 28,
5
- "a": 23,
6
- "b": 0,
7
- "c": 16,
8
- "d": 21,
9
- "e": 5,
10
- "f": 24,
11
- "g": 12,
12
- "h": 6,
13
- "i": 27,
14
- "j": 26,
15
- "k": 17,
16
- "l": 19,
17
- "m": 15,
18
- "n": 4,
19
- "o": 10,
20
- "p": 18,
21
  "q": 1,
22
- "r": 7,
23
- "s": 8,
24
- "t": 13,
25
- "u": 11,
26
- "v": 25,
27
- "w": 14,
28
- "x": 22,
29
- "y": 3,
30
- "z": 2,
31
- "|": 20
32
  }
 
1
  {
2
+ "'": 6,
3
  "[PAD]": 29,
4
  "[UNK]": 28,
5
+ "a": 4,
6
+ "b": 17,
7
+ "c": 20,
8
+ "d": 2,
9
+ "e": 11,
10
+ "f": 27,
11
+ "g": 9,
12
+ "h": 23,
13
+ "i": 5,
14
+ "j": 7,
15
+ "k": 0,
16
+ "l": 10,
17
+ "m": 19,
18
+ "n": 13,
19
+ "o": 22,
20
+ "p": 3,
21
  "q": 1,
22
+ "r": 14,
23
+ "s": 24,
24
+ "t": 16,
25
+ "u": 12,
26
+ "v": 8,
27
+ "w": 21,
28
+ "x": 25,
29
+ "y": 26,
30
+ "z": 18,
31
+ "|": 15
32
  }