File size: 1,802 Bytes
5012e22
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d2f4d5b
 
 
5012e22
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d2f4d5b
5012e22
d2f4d5b
5012e22
d2f4d5b
 
5012e22
 
 
 
 
 
 
 
 
 
d2f4d5b
 
 
 
 
5012e22
 
 
 
 
 
 
 
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
---
license: apache-2.0
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: openai/whisper-medium
  results: []
---

<!-- 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. -->

# openai/whisper-medium

This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7427
- Wer: 19.3902
- Cer: 8.7285

## 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: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 5000
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Wer     | Cer    |
|:-------------:|:-----:|:----:|:---------------:|:-------:|:------:|
| 0.0471        | 1.05  | 1000 | 0.5961          | 20.2895 | 8.9820 |
| 0.0194        | 2.11  | 2000 | 1.0999          | 22.6146 | 9.7105 |
| 0.002         | 4.02  | 3000 | 0.7289          | 20.2018 | 8.8379 |
| 0.0006        | 5.07  | 4000 | 0.7791          | 19.3683 | 8.4376 |
| 0.0003        | 6.13  | 5000 | 0.7427          | 19.3902 | 8.7285 |


### Framework versions

- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu117
- Datasets 2.7.1.dev0
- Tokenizers 0.13.2