File size: 2,265 Bytes
7884c85
 
 
 
 
 
 
 
 
671c3dc
 
7884c85
 
671c3dc
 
 
 
 
 
 
 
 
 
 
 
 
 
7884c85
 
 
 
 
 
 
 
671c3dc
 
 
7884c85
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
671c3dc
7884c85
 
671c3dc
7884c85
 
 
 
671c3dc
7884c85
 
 
 
671c3dc
 
 
 
 
 
 
 
 
 
7884c85
 
 
 
 
 
 
 
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
89
---
language:
- en
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- librispeech_asr
metrics:
- wer
model-index:
- name: SpeechGPT
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: librispeech_asr
      type: librispeech_asr
      config: clean
      split: None
      args: 'config: clean, split: train'
    metrics:
    - name: Wer
      type: wer
      value: 23.544963481436394
---

<!-- 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. -->

# SpeechGPT

This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the librispeech_asr dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5354
- Wer: 23.5450

## 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: 500
- num_epochs: 1
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Wer     |
|:-------------:|:-----:|:----:|:---------------:|:-------:|
| 1.0518        | 0.12  | 1000 | 0.7491          | 31.7103 |
| 0.8884        | 0.24  | 2000 | 0.6588          | 27.4003 |
| 0.8061        | 0.36  | 3000 | 0.6177          | 26.4569 |
| 0.8549        | 0.48  | 4000 | 0.5888          | 25.5002 |
| 0.7836        | 0.6   | 5000 | 0.5688          | 25.8939 |
| 0.691         | 0.72  | 6000 | 0.5542          | 24.1574 |
| 0.7044        | 0.84  | 7000 | 0.5429          | 23.5450 |
| 0.7309        | 0.97  | 8000 | 0.5354          | 23.5450 |


### Framework versions

- Transformers 4.39.3
- Pytorch 2.2.2+cu121
- Datasets 2.19.0
- Tokenizers 0.15.2