File size: 2,417 Bytes
c53d1f1
 
 
 
 
 
 
 
 
 
 
 
fcd8937
c53d1f1
 
 
 
 
 
 
 
 
 
 
 
 
fcd8937
c53d1f1
 
 
 
 
fcd8937
c53d1f1
 
 
fcd8937
 
 
c53d1f1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fcd8937
c53d1f1
 
 
 
 
 
fcd8937
 
 
 
 
 
 
 
c53d1f1
 
 
 
 
 
 
 
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
83
84
85
86
87
88
---
language:
- ru
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- mozilla-foundation/common_voice_13_0
metrics:
- wer
model-index:
- name: Whisper Small Ru - Model_ru_3
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: Common Voice 13
      type: mozilla-foundation/common_voice_13_0
      config: ru
      split: test
      args: ru
    metrics:
    - name: Wer
      type: wer
      value: 13.30140186915888
---

<!-- 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 Ru - Model_ru_3

This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Common Voice 13 dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2080
- Wer Ortho: 17.4462
- Wer: 13.3014

## 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: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 50
- training_steps: 4000
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer     |
|:-------------:|:-----:|:----:|:---------------:|:---------:|:-------:|
| 0.2085        | 0.22  | 500  | 0.2366          | 19.9234   | 14.9498 |
| 0.1875        | 0.44  | 1000 | 0.2176          | 19.3079   | 14.5643 |
| 0.1688        | 0.66  | 1500 | 0.2095          | 18.3736   | 13.9287 |
| 0.1678        | 0.88  | 2000 | 0.2038          | 17.7325   | 13.4381 |
| 0.0853        | 1.1   | 2500 | 0.2036          | 17.0309   | 12.7488 |
| 0.0822        | 1.32  | 3000 | 0.2046          | 17.6894   | 13.2780 |
| 0.0775        | 1.54  | 3500 | 0.2051          | 16.9948   | 12.7126 |
| 0.0727        | 1.76  | 4000 | 0.2080          | 17.4462   | 13.3014 |


### Framework versions

- Transformers 4.38.2
- Pytorch 2.3.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2