Hugging Face's logo

dmusingu
/
luganda_wav2vec2_ctc_tokenizer

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Eval Results (legacy)
Model card Files
xet
Metrics Community
dmusingu commited on
Commit
ac0c265
·
verified ·
1 Parent(s): f74ff30

Upload tokenizer

Browse files
Files changed (1) hide show
  1. vocab.json +29 -29
vocab.json CHANGED
@@ -1,35 +1,35 @@
1
  {
2
- "'": 18,
3
- "(": 12,
4
- ")": 0,
5
  "[PAD]": 32,
6
  "[UNK]": 31,
7
- "a": 23,
8
- "b": 24,
9
- "c": 13,
10
- "d": 21,
11
- "e": 11,
12
- "f": 9,
13
- "g": 1,
14
- "h": 27,
15
- "i": 17,
16
- "j": 29,
17
- "k": 14,
18
- "l": 2,
19
- "m": 22,
20
- "n": 10,
21
- "o": 15,
22
- "p": 6,
23
  "r": 28,
24
  "s": 30,
25
- "t": 26,
26
- "u": 8,
27
- "v": 25,
28
- "w": 3,
29
- "x": 19,
30
- "y": 16,
31
- "z": 20,
32
- "|": 7,
33
- "‘": 4,
34
- "’": 5
35
  }
 
1
  {
2
+ "'": 23,
3
+ "(": 22,
4
+ ")": 1,
5
  "[PAD]": 32,
6
  "[UNK]": 31,
7
+ "a": 11,
8
+ "b": 15,
9
+ "c": 14,
10
+ "d": 4,
11
+ "e": 19,
12
+ "f": 17,
13
+ "g": 24,
14
+ "h": 16,
15
+ "i": 20,
16
+ "j": 2,
17
+ "k": 5,
18
+ "l": 13,
19
+ "m": 21,
20
+ "n": 7,
21
+ "o": 27,
22
+ "p": 8,
23
  "r": 28,
24
  "s": 30,
25
+ "t": 29,
26
+ "u": 0,
27
+ "v": 10,
28
+ "w": 9,
29
+ "x": 25,
30
+ "y": 18,
31
+ "z": 3,
32
+ "|": 12,
33
+ "‘": 26,
34
+ "’": 6
35
  }