Hugging Face's logo

DewiBrynJones
/
whisper-large-v3-turbo-ft-cv-cy-train-all-plus-other-with-excluded

TensorBoard
Safetensors
whisper
Generated from Trainer
Model card Files Metrics Community
DewiBrynJones commited on
Commit
eda2019
•
1 Parent(s): 796e523

Model save

Browse files
Files changed (2) hide show
  1. README.md +10 -10
  2. generation_config.json +11 -27
README.md CHANGED
@@ -1,6 +1,6 @@
1
  ---
2
- license: apache-2.0
3
- base_model: openai/whisper-large-v3
4
  tags:
5
  - generated_from_trainer
6
  metrics:
@@ -15,10 +15,10 @@ should probably proofread and complete it, then remove this comment. -->
15
 
16
  # whisper-large-v3-turbo-ft-cv-cy-train-all-plus-other-with-excluded
17
 
18
- This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the DewiBrynJones/commonvoice_18_0_cy train_all+other_with_excluded main dataset.
19
  It achieves the following results on the evaluation set:
20
- - Loss: 0.3181
21
- - Wer: 0.1683
22
 
23
  ## Model description
24
 
@@ -53,11 +53,11 @@ The following hyperparameters were used during training:
53
 
54
  | Training Loss | Epoch | Step | Validation Loss | Wer |
55
  |:-------------:|:------:|:----:|:---------------:|:------:|
56
- | 0.1599 | 1.4144 | 1000 | 0.2516 | 0.2035 |
57
- | 0.0651 | 2.8289 | 2000 | 0.2339 | 0.1831 |
58
- | 0.0119 | 4.2433 | 3000 | 0.2720 | 0.1751 |
59
- | 0.005 | 5.6577 | 4000 | 0.2928 | 0.1692 |
60
- | 0.0009 | 7.0721 | 5000 | 0.3181 | 0.1683 |
61
 
62
 
63
  ### Framework versions
 
1
  ---
2
+ license: mit
3
+ base_model: openai/whisper-large-v3-turbo
4
  tags:
5
  - generated_from_trainer
6
  metrics:
 
15
 
16
  # whisper-large-v3-turbo-ft-cv-cy-train-all-plus-other-with-excluded
17
 
18
+ This model is a fine-tuned version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) on an unknown dataset.
19
  It achieves the following results on the evaluation set:
20
+ - Loss: 0.3134
21
+ - Wer: 0.1746
22
 
23
  ## Model description
24
 
 
53
 
54
  | Training Loss | Epoch | Step | Validation Loss | Wer |
55
  |:-------------:|:------:|:----:|:---------------:|:------:|
56
+ | 0.2147 | 1.4144 | 1000 | 0.3066 | 0.2349 |
57
+ | 0.0989 | 2.8289 | 2000 | 0.2775 | 0.2072 |
58
+ | 0.0295 | 4.2433 | 3000 | 0.2935 | 0.1919 |
59
+ | 0.0109 | 5.6577 | 4000 | 0.3011 | 0.1828 |
60
+ | 0.0016 | 7.0721 | 5000 | 0.3134 | 0.1746 |
61
 
62
 
63
  ### Framework versions
generation_config.json CHANGED
@@ -1,44 +1,28 @@
1
  {
2
  "alignment_heads": [
3
  [
4
- 7,
5
- 0
6
- ],
7
- [
8
- 10,
9
- 17
10
- ],
11
- [
12
- 12,
13
- 18
14
  ],
15
  [
16
- 13,
17
- 12
18
  ],
19
  [
20
- 16,
21
- 1
22
  ],
23
  [
24
- 17,
25
- 14
26
  ],
27
  [
28
- 19,
29
  11
30
  ],
31
  [
32
- 21,
33
- 4
34
- ],
35
- [
36
- 24,
37
- 1
38
- ],
39
- [
40
- 25,
41
- 6
42
  ]
43
  ],
44
  "begin_suppress_tokens": [
 
1
  {
2
  "alignment_heads": [
3
  [
4
+ 2,
5
+ 4
 
 
 
 
 
 
 
 
6
  ],
7
  [
8
+ 2,
9
+ 11
10
  ],
11
  [
12
+ 3,
13
+ 3
14
  ],
15
  [
16
+ 3,
17
+ 6
18
  ],
19
  [
20
+ 3,
21
  11
22
  ],
23
  [
24
+ 3,
25
+ 14
 
 
 
 
 
 
 
 
26
  ]
27
  ],
28
  "begin_suppress_tokens": [