Mihaj commited on
Commit
bbfd945
1 Parent(s): c2622a7

Upload tokenizer

Browse files
Files changed (2) hide show
  1. tokenizer_config.json +0 -1
  2. vocab.json +28 -28
tokenizer_config.json CHANGED
@@ -39,7 +39,6 @@
39
  "eos_token": "</s>",
40
  "model_max_length": 1000000000000000019884624838656,
41
  "pad_token": "[PAD]",
42
- "processor_class": "Wav2Vec2BertProcessor",
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
- "'": 24,
3
- "a": 14,
4
- "b": 12,
5
- "c": 0,
6
- "d": 11,
7
- "e": 17,
8
- "f": 9,
9
- "g": 15,
10
- "h": 18,
11
- "i": 6,
12
- "j": 22,
13
- "k": 27,
14
- "l": 20,
15
- "m": 10,
16
- "n": 5,
17
- "o": 26,
18
- "p": 19,
19
- "r": 16,
20
- "s": 21,
21
- "t": 29,
22
- "u": 4,
23
- "v": 7,
24
- "y": 28,
25
- "z": 23,
26
- "|": 3,
27
  "ä": 2,
28
- "ö": 13,
29
- "č": 1,
30
- "š": 25,
31
  "ž": 8
32
  }
 
1
  {
2
+ "'": 13,
3
+ "a": 21,
4
+ "b": 25,
5
+ "c": 14,
6
+ "d": 3,
7
+ "e": 26,
8
+ "f": 11,
9
+ "g": 27,
10
+ "h": 9,
11
+ "i": 5,
12
+ "j": 10,
13
+ "k": 15,
14
+ "l": 24,
15
+ "m": 18,
16
+ "n": 7,
17
+ "o": 23,
18
+ "p": 1,
19
+ "r": 4,
20
+ "s": 12,
21
+ "t": 19,
22
+ "u": 22,
23
+ "v": 17,
24
+ "y": 20,
25
+ "z": 28,
26
+ "|": 16,
27
  "ä": 2,
28
+ "ö": 29,
29
+ "č": 0,
30
+ "š": 6,
31
  "ž": 8
32
  }