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
cc928d9
•
1 Parent(s): ec27ecb

Upload tokenizer

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