Reihaneh commited on
Commit
02779ef
·
1 Parent(s): 3cb44ec

Upload tokenizer

Browse files
Files changed (2) hide show
  1. tokenizer_config.json +1 -0
  2. vocab.json +37 -37
tokenizer_config.json CHANGED
@@ -6,6 +6,7 @@
6
  "model_max_length": 1000000000000000019884624838656,
7
  "pad_token": "[PAD]",
8
  "replace_word_delimiter_char": " ",
 
9
  "tokenizer_class": "Wav2Vec2CTCTokenizer",
10
  "unk_token": "[UNK]",
11
  "word_delimiter_token": "|"
 
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
@@ -1,42 +1,42 @@
1
  {
2
- "'": 35,
3
  "[PAD]": 39,
4
  "[UNK]": 38,
5
- "a": 16,
6
- "b": 24,
7
- "c": 8,
8
- "d": 11,
9
- "e": 30,
10
- "f": 7,
11
- "g": 12,
12
- "h": 14,
13
- "i": 5,
14
- "j": 4,
15
- "k": 34,
16
- "l": 25,
17
- "m": 10,
18
- "n": 9,
19
- "o": 22,
20
- "p": 18,
21
- "r": 3,
22
- "s": 37,
23
- "t": 33,
24
- "u": 1,
25
- "v": 23,
26
- "w": 2,
27
- "y": 6,
28
- "z": 0,
29
- "|": 19,
30
- "á": 13,
31
- "â": 21,
32
- "ä": 31,
33
- "è": 26,
34
- "é": 28,
35
- "ê": 15,
36
- "ë": 17,
37
  "ï": 32,
38
- "ô": 29,
39
- "ú": 20,
40
- "û": 27,
41
- "ü": 36
42
  }
 
1
  {
2
+ "'": 33,
3
  "[PAD]": 39,
4
  "[UNK]": 38,
5
+ "a": 25,
6
+ "b": 23,
7
+ "c": 30,
8
+ "d": 37,
9
+ "e": 31,
10
+ "f": 8,
11
+ "g": 1,
12
+ "h": 15,
13
+ "i": 7,
14
+ "j": 35,
15
+ "k": 12,
16
+ "l": 24,
17
+ "m": 34,
18
+ "n": 27,
19
+ "o": 21,
20
+ "p": 6,
21
+ "r": 36,
22
+ "s": 9,
23
+ "t": 18,
24
+ "u": 10,
25
+ "v": 5,
26
+ "w": 13,
27
+ "y": 11,
28
+ "z": 3,
29
+ "|": 0,
30
+ "á": 22,
31
+ "â": 19,
32
+ "ä": 20,
33
+ "è": 4,
34
+ "é": 2,
35
+ "ê": 14,
36
+ "ë": 28,
37
  "ï": 32,
38
+ "ô": 26,
39
+ "ú": 29,
40
+ "û": 17,
41
+ "ü": 16
42
  }