Hugging Face's logo

akera
/
mms-en

Model card Files Community
akera commited on
Commit
1d7c419
•
1 Parent(s): 1de9715

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +1 -1
  3. vocab.json +39 -33
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 33,
3
- "<s>": 32
4
  }
 
1
  {
2
+ "</s>": 39,
3
+ "<s>": 38
4
  }
tokenizer_config.json CHANGED
@@ -6,7 +6,7 @@
6
  "model_max_length": 1000000000000000019884624838656,
7
  "pad_token": "[PAD]",
8
  "replace_word_delimiter_char": " ",
9
- "target_lang": "ach",
10
  "tokenizer_class": "Wav2Vec2CTCTokenizer",
11
  "unk_token": "[UNK]",
12
  "word_delimiter_token": "|"
 
6
  "model_max_length": 1000000000000000019884624838656,
7
  "pad_token": "[PAD]",
8
  "replace_word_delimiter_char": " ",
9
+ "target_lang": "en",
10
  "tokenizer_class": "Wav2Vec2CTCTokenizer",
11
  "unk_token": "[UNK]",
12
  "word_delimiter_token": "|"
vocab.json CHANGED
@@ -1,36 +1,42 @@
1
  {
2
- "ach": {
3
- "=": 1,
4
- "[": 2,
5
- "[PAD]": 31,
6
- "[UNK]": 30,
7
- "]": 3,
8
- "a": 4,
9
- "b": 5,
10
- "c": 6,
11
- "d": 7,
12
- "e": 8,
13
- "f": 9,
14
- "g": 10,
15
- "h": 11,
16
- "i": 12,
17
- "j": 13,
18
- "k": 14,
19
- "l": 15,
20
- "m": 16,
21
- "n": 17,
22
- "o": 18,
23
- "p": 19,
24
- "q": 20,
25
- "r": 21,
26
- "s": 22,
27
- "t": 23,
28
- "u": 24,
29
- "v": 25,
30
- "w": 26,
31
- "x": 27,
32
- "y": 28,
33
- "z": 29,
34
- "|": 0
 
 
 
 
 
 
35
  }
36
  }
 
1
  {
2
+ "en": {
3
+ ">": 1,
4
+ "[PAD]": 37,
5
+ "[UNK]": 36,
6
+ "`": 2,
7
+ "a": 3,
8
+ "b": 4,
9
+ "c": 5,
10
+ "d": 6,
11
+ "e": 7,
12
+ "f": 8,
13
+ "g": 9,
14
+ "h": 10,
15
+ "i": 11,
16
+ "j": 12,
17
+ "k": 13,
18
+ "l": 14,
19
+ "m": 15,
20
+ "n": 16,
21
+ "o": 17,
22
+ "p": 18,
23
+ "q": 19,
24
+ "r": 20,
25
+ "s": 21,
26
+ "t": 22,
27
+ "u": 23,
28
+ "v": 24,
29
+ "w": 25,
30
+ "x": 26,
31
+ "y": 27,
32
+ "z": 28,
33
+ "|": 0,
34
+ "´": 29,
35
+ "ä": 30,
36
+ "é": 31,
37
+ "ô": 32,
38
+ "‚": 33,
39
+ "“": 34,
40
+ "”": 35
41
  }
42
  }