File size: 2,142 Bytes
e12b4f3
 
 
 
 
 
 
fca84d8
e12b4f3
 
 
fca84d8
e12b4f3
 
 
 
 
fca84d8
 
e12b4f3
 
fca84d8
e12b4f3
 
 
7d17059
e12b4f3
 
 
 
 
fca84d8
e12b4f3
fca84d8
e12b4f3
7d17059
 
e12b4f3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7d17059
 
 
 
e12b4f3
 
 
 
 
 
 
 
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
77
78
79
80
81
82
---

library_name: transformers
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- common_voice_11_0
metrics:
- wer
model-index:
- name: whisper-small-chinese-tw
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: common_voice_11_0
      type: common_voice_11_0
      config: zh-TW
      split: test
      args: zh-TW
    metrics:
    - name: Wer
      type: wer
      value: 39.73242726693565
---


<!-- 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-chinese-tw

This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the common_voice_11_0 dataset.

It achieves the following results on the evaluation set:

- Loss: 0.2167

- Wer: 39.7324



## 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
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- 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     |
|:-------------:|:------:|:----:|:---------------:|:-------:|
| 0.0957        | 1.4184 | 1000 | 0.1986          | 40.8792 |
| 0.0315        | 2.8369 | 2000 | 0.2038          | 40.4544 |
| 0.0036        | 4.2553 | 3000 | 0.2102          | 40.1571 |
| 0.0018        | 5.6738 | 4000 | 0.2167          | 39.7324 |


### Framework versions

- Transformers 4.46.2
- Pytorch 2.4.0+cu124
- Datasets 3.1.0
- Tokenizers 0.20.3