Hugging Face's logo

Rziane
/
wav2vec2-large-mms-1b-tun_test

Safetensors
wav2vec2
Generated from Trainer
Model card Files Community
Rziane commited on
Commit
98638a4
·
verified ·
1 Parent(s): 5706b89

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +42 -45
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 58,
3
- "<s>": 57
4
  }
 
1
  {
2
+ "</s>": 55,
3
+ "<s>": 54
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "55": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "56": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "57": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "58": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "52": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "53": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "54": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "55": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -11,51 +11,48 @@
11
  "7": 9,
12
  "8": 10,
13
  "9": 11,
14
- "[PAD]": 56,
15
- "[UNK]": 55,
16
- "^": 12,
17
- "a": 13,
18
- "b": 14,
19
- "c": 15,
20
- "d": 16,
21
- "e": 17,
22
- "f": 18,
23
- "g": 19,
24
- "h": 20,
25
- "i": 21,
26
- "j": 22,
27
- "k": 23,
28
- "l": 24,
29
- "m": 25,
30
- "n": 26,
31
- "o": 27,
32
- "p": 28,
33
- "q": 29,
34
- "r": 30,
35
- "s": 31,
36
- "t": 32,
37
- "u": 33,
38
- "v": 34,
39
- "w": 35,
40
- "x": 36,
41
- "y": 37,
42
- "z": 38,
43
- "{": 39,
44
  "|": 0,
45
- "}": 40,
46
- "à": 41,
47
- "ç": 42,
48
- "è": 43,
49
- "é": 44,
50
- "ê": 45,
51
- "ù": 46,
52
- "ď": 47,
53
- "đ": 48,
54
- "ġ": 49,
55
- "š": 50,
56
- "ŧ": 51,
57
- "ž": 52,
58
- "ԑ": 53,
59
- "ḥ": 54
60
  }
61
  }
 
11
  "7": 9,
12
  "8": 10,
13
  "9": 11,
14
+ "[PAD]": 53,
15
+ "[UNK]": 52,
16
+ "a": 12,
17
+ "b": 13,
18
+ "c": 14,
19
+ "d": 15,
20
+ "e": 16,
21
+ "f": 17,
22
+ "g": 18,
23
+ "h": 19,
24
+ "i": 20,
25
+ "j": 21,
26
+ "k": 22,
27
+ "l": 23,
28
+ "m": 24,
29
+ "n": 25,
30
+ "o": 26,
31
+ "p": 27,
32
+ "q": 28,
33
+ "r": 29,
34
+ "s": 30,
35
+ "t": 31,
36
+ "u": 32,
37
+ "v": 33,
38
+ "w": 34,
39
+ "x": 35,
40
+ "y": 36,
41
+ "z": 37,
 
 
42
  "|": 0,
43
+ "à": 38,
44
+ "ç": 39,
45
+ "è": 40,
46
+ "é": 41,
47
+ "ê": 42,
48
+ "ù": 43,
49
+ "ď": 44,
50
+ "đ": 45,
51
+ "ġ": 46,
52
+ "š": 47,
53
+ "ŧ": 48,
54
+ "ž": 49,
55
+ "ԑ": 50,
56
+ "ḥ": 51
 
57
  }
58
  }