Hugging Face's logo

khaoulaoub
/
wav2vec2_XLSR_darija_maroc

Automatic Speech Recognition
Transformers
PyTorch
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Community
1
khaoulaoub commited on
Commit
7015e69
•
1 Parent(s): 180d66b

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +24 -23
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 35,
3
- "<s>": 34
4
  }
 
1
  {
2
+ "</s>": 36,
3
+ "<s>": 35
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "32": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "33": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "34": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": true,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "35": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": true,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "33": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "34": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "35": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": true,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "36": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": true,
vocab.json CHANGED
@@ -1,36 +1,37 @@
1
  {
2
- "[PAD]": 33,
3
- "[UNK]": 32,
4
- "|": 10,
5
- "ء": 11,
6
  "ا": 7,
7
- "ب": 22,
8
- "ة": 29,
9
- "ت": 26,
10
- "ث": 15,
11
- "ج": 30,
12
- "ح": 18,
13
- "خ": 14,
14
  "د": 8,
15
  "ذ": 24,
16
  "ر": 2,
17
  "ز": 1,
18
- "س": 19,
19
- "ش": 16,
20
  "ص": 5,
21
  "ض": 4,
22
- "ط": 25,
23
  "ظ": 6,
24
  "ع": 13,
25
- "غ": 23,
26
- "ف": 12,
27
- "ق": 31,
28
- "ك": 27,
 
29
  "ل": 3,
30
  "م": 0,
31
- "ن": 17,
32
- "ه": 20,
33
- "و": 28,
34
- "ى": 9,
35
- "ي": 21
36
  }
 
1
  {
2
+ "[PAD]": 34,
3
+ "[UNK]": 33,
4
+ "|": 11,
5
+ "ء": 12,
6
  "ا": 7,
7
+ "ب": 23,
8
+ "ة": 30,
9
+ "ت": 27,
10
+ "ث": 16,
11
+ "ج": 31,
12
+ "ح": 19,
13
+ "خ": 15,
14
  "د": 8,
15
  "ذ": 24,
16
  "ر": 2,
17
  "ز": 1,
18
+ "س": 20,
19
+ "ش": 17,
20
  "ص": 5,
21
  "ض": 4,
22
+ "ط": 26,
23
  "ظ": 6,
24
  "ع": 13,
25
+ "غ": 25,
26
+ "ـ": 9,
27
+ "ف": 14,
28
+ "ق": 32,
29
+ "ك": 28,
30
  "ل": 3,
31
  "م": 0,
32
+ "ن": 18,
33
+ "ه": 21,
34
+ "و": 29,
35
+ "ى": 10,
36
+ "ي": 22
37
  }