File size: 1,868 Bytes
57280ea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
---
license: apache-2.0
base_model: nikes64/whisper-small-uk
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper-small-uk-v2
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    metrics:
    - name: Wer
      type: wer
      value: 11.2911
language:
- uk
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# whisper-small-uk-v2

This model is a fine-tuned version of [nikes64/whisper-small-uk](https://huggingface.co/nikes64/whisper-small-uk) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1369
- Wer: 11.2911

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 4000
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Wer     |
|:-------------:|:-----:|:----:|:---------------:|:-------:|
| 0.087         | 1.4   | 1000 | 0.1092          | 11.8439 |
| 0.0429        | 2.8   | 2000 | 0.1035          | 11.0597 |
| 0.0132        | 4.2   | 3000 | 0.1247          | 11.1431 |
| 0.0067        | 5.6   | 4000 | 0.1369          | 11.2911 |


### Framework versions

- Transformers 4.38.0.dev0
- Pytorch 2.2.0+cu121
- Datasets 2.17.1
- Tokenizers 0.15.2