Hugging Face's logo

eddiegulay
/
wav2vec2-large-xlsr-mvc-swahili

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
Swahili
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
eddiegulay commited on
Commit
11635f3
•
1 Parent(s): fd84a65

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +31 -32
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 34,
3
- "<s>": 33
4
  }
 
1
  {
2
+ "</s>": 33,
3
+ "<s>": 32
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "31": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "32": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "33": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": true,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "34": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": true,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "30": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "31": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "32": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": true,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "33": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": true,
vocab.json CHANGED
@@ -1,35 +1,34 @@
1
  {
2
- "[PAD]": 32,
3
- "[UNK]": 31,
4
- "_": 9,
5
- "a": 15,
6
- "b": 24,
7
- "c": 21,
8
- "d": 20,
9
- "e": 30,
10
- "f": 23,
11
- "g": 12,
12
  "h": 13,
13
- "i": 14,
14
- "j": 5,
15
- "k": 22,
16
- "l": 16,
17
- "m": 29,
18
- "n": 7,
19
- "o": 8,
20
- "p": 28,
21
- "q": 18,
22
- "r": 1,
23
- "s": 6,
24
- "t": 2,
25
- "u": 10,
26
- "v": 17,
27
- "w": 11,
28
- "x": 25,
29
- "y": 19,
30
- "z": 4,
31
- "|": 26,
32
- "ã": 27,
33
- "å": 3,
34
- "š": 0
35
  }
 
1
  {
2
+ "[PAD]": 31,
3
+ "[UNK]": 30,
4
+ "a": 10,
5
+ "b": 27,
6
+ "c": 16,
7
+ "d": 15,
8
+ "e": 21,
9
+ "f": 8,
10
+ "g": 24,
 
11
  "h": 13,
12
+ "i": 26,
13
+ "j": 4,
14
+ "k": 17,
15
+ "l": 12,
16
+ "m": 1,
17
+ "n": 20,
18
+ "o": 0,
19
+ "p": 23,
20
+ "q": 29,
21
+ "r": 3,
22
+ "s": 9,
23
+ "t": 22,
24
+ "u": 2,
25
+ "v": 11,
26
+ "w": 25,
27
+ "x": 28,
28
+ "y": 18,
29
+ "z": 14,
30
+ "|": 6,
31
+ "ã": 5,
32
+ "å": 19,
33
+ "š": 7
34
  }