File size: 2,182 Bytes
6f2a080
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
language:
- ru
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- tbkazakova/even_speech_biblical
metrics:
- wer
model-index:
- name: Whisper Small Even - VovaK13
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: Even Speech Biblical
      type: tbkazakova/even_speech_biblical
      config: default
      split: None
      args: 'config: ru, split: train'
    metrics:
    - name: Wer
      type: wer
      value: 30.275689223057643
---

<!-- 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 Even - VovaK13

This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Even Speech Biblical dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4483
- Wer: 30.2757

## 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: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 250
- training_steps: 2500
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch   | Step | Validation Loss | Wer     |
|:-------------:|:-------:|:----:|:---------------:|:-------:|
| 0.0509        | 5.9880  | 500  | 0.3699          | 33.7343 |
| 0.0022        | 11.9760 | 1000 | 0.4084          | 30.9273 |
| 0.0003        | 17.9641 | 1500 | 0.4336          | 30.1253 |
| 0.0002        | 23.9521 | 2000 | 0.4444          | 30.2757 |
| 0.0002        | 29.9401 | 2500 | 0.4483          | 30.2757 |


### Framework versions

- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1