Hugging Face's logo

Jubliano
/
wav2vec2-large-xls-r-300m-ipa

Automatic Speech Recognition
Transformers
Safetensors
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Community
Jubliano commited on
Commit
d9cf2f0
•
1 Parent(s): fe04a40

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +22 -3
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 115,
3
- "<s>": 114
4
  }
 
1
  {
2
+ "</s>": 134,
3
+ "<s>": 133
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "112": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "113": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "114": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "115": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "131": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "132": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "133": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "134": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -7,8 +7,8 @@
7
  "4": 109,
8
  "5": 104,
9
  "6": 105,
10
- "[PAD]": 113,
11
- "[UNK]": 112,
12
  "a": 84,
13
  "b": 1,
14
  "c": 6,
@@ -35,7 +35,7 @@
35
  "x": 41,
36
  "y": 61,
37
  "z": 34,
38
- "|": 111,
39
  "ä": 98,
40
  "æ": 82,
41
  "ç": 39,
@@ -108,6 +108,25 @@
108
  "ː": 97,
109
  "ˠ": 91,
110
  "ˤ": 92,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
111
  "β": 28,
112
  "θ": 31,
113
  "χ": 43,
 
7
  "4": 109,
8
  "5": 104,
9
  "6": 105,
10
+ "[PAD]": 132,
11
+ "[UNK]": 131,
12
  "a": 84,
13
  "b": 1,
14
  "c": 6,
 
35
  "x": 41,
36
  "y": 61,
37
  "z": 34,
38
+ "|": 130,
39
  "ä": 98,
40
  "æ": 82,
41
  "ç": 39,
 
108
  "ː": 97,
109
  "ˠ": 91,
110
  "ˤ": 92,
111
+ "̃": 129,
112
+ "̈": 118,
113
+ "̊": 112,
114
+ "̘": 124,
115
+ "̙": 125,
116
+ "̜": 115,
117
+ "̝": 123,
118
+ "̞": 122,
119
+ "̟": 116,
120
+ "̠": 117,
121
+ "̤": 126,
122
+ "̥": 111,
123
+ "̩": 120,
124
+ "̬": 113,
125
+ "̯": 121,
126
+ "̰": 127,
127
+ "̹": 114,
128
+ "̼": 128,
129
+ "̽": 119,
130
  "β": 28,
131
  "θ": 31,
132
  "χ": 43,