Hugging Face's logo

hilaltekgoz
/
wav2vec2-large-mms-1b-turkish-colab

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
hilaltekgoz commited on
Commit
0e96a49
·
verified ·
1 Parent(s): 1cfe4c4

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +38 -39
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 41,
3
- "<s>": 40
4
  }
 
1
  {
2
+ "</s>": 40,
3
+ "<s>": 39
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "38": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "39": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "40": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "41": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "37": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "38": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "39": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "40": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,44 +1,43 @@
1
  {
2
  "tur": {
3
- "[PAD]": 39,
4
- "[UNK]": 38,
5
- "a": 1,
6
- "b": 2,
7
- "c": 3,
8
- "d": 4,
9
- "e": 5,
10
- "f": 6,
11
- "g": 7,
12
- "h": 8,
13
- "i": 9,
14
- "j": 10,
15
- "k": 11,
16
- "l": 12,
17
- "m": 13,
18
- "n": 14,
19
- "o": 15,
20
- "p": 16,
21
- "q": 17,
22
- "r": 18,
23
- "s": 19,
24
- "t": 20,
25
- "u": 21,
26
- "v": 22,
27
- "w": 23,
28
- "x": 24,
29
- "y": 25,
30
- "z": 26,
 
31
  "|": 0,
32
- "ç": 27,
33
- "é": 28,
34
- "ë": 29,
35
- "ö": 30,
36
- "ü": 31,
37
- "ğ": 32,
38
- "ı": 33,
39
- "ş": 34,
40
- "̇": 35,
41
- "’": 36,
42
- "…": 37
43
  }
44
  }
 
1
  {
2
  "tur": {
3
+ "(": 1,
4
+ ")": 2,
5
+ "[PAD]": 38,
6
+ "[UNK]": 37,
7
+ "a": 3,
8
+ "b": 4,
9
+ "c": 5,
10
+ "d": 6,
11
+ "e": 7,
12
+ "f": 8,
13
+ "g": 9,
14
+ "h": 10,
15
+ "i": 11,
16
+ "j": 12,
17
+ "k": 13,
18
+ "l": 14,
19
+ "m": 15,
20
+ "n": 16,
21
+ "o": 17,
22
+ "p": 18,
23
+ "r": 19,
24
+ "s": 20,
25
+ "t": 21,
26
+ "u": 22,
27
+ "v": 23,
28
+ "w": 24,
29
+ "x": 25,
30
+ "y": 26,
31
+ "z": 27,
32
  "|": 0,
33
+ "ç": 28,
34
+ "ö": 29,
35
+ "ü": 30,
36
+ "ğ": 31,
37
+ "ı": 32,
38
+ "ş": 33,
39
+ "̇": 34,
40
+ "’": 35,
41
+ "…": 36
 
 
42
  }
43
  }