File size: 2,513 Bytes
0bbdad1
 
 
 
 
bea1a8e
0bbdad1
bea1a8e
 
0bbdad1
 
 
 
 
bea1a8e
0bbdad1
399d0d4
0bbdad1
399d0d4
 
0bbdad1
 
 
bea1a8e
0bbdad1
 
 
bea1a8e
0bbdad1
 
 
bea1a8e
0bbdad1
 
 
 
 
 
89dcecc
399d0d4
0bbdad1
 
 
399d0d4
0bbdad1
 
 
c8b6183
0bbdad1
 
 
 
399d0d4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0bbdad1
 
 
 
bea1a8e
 
 
0bbdad1
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
---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- common_voice
model-index:
- name: xls-r-uyghur-cv8
  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. -->

# xls-r-uyghur-cv8

This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the common_voice dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2024
- Wer: 0.3280

## 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: 0.0001
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- 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: 2000
- num_epochs: 100.0
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Wer    |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| 3.3036        | 5.32  | 500  | 3.2628          | 1.0    |
| 2.9734        | 10.63 | 1000 | 2.5677          | 0.9980 |
| 1.3466        | 15.95 | 1500 | 0.4455          | 0.6306 |
| 1.2424        | 21.28 | 2000 | 0.3603          | 0.5301 |
| 1.1655        | 26.59 | 2500 | 0.3165          | 0.4740 |
| 1.1026        | 31.91 | 3000 | 0.2930          | 0.4400 |
| 1.0655        | 37.23 | 3500 | 0.2675          | 0.4159 |
| 1.0239        | 42.55 | 4000 | 0.2580          | 0.3913 |
| 0.9938        | 47.87 | 4500 | 0.2373          | 0.3698 |
| 0.9655        | 53.19 | 5000 | 0.2379          | 0.3675 |
| 0.9374        | 58.51 | 5500 | 0.2486          | 0.3795 |
| 0.9065        | 63.83 | 6000 | 0.2243          | 0.3405 |
| 0.888         | 69.15 | 6500 | 0.2157          | 0.3277 |
| 0.8646        | 74.47 | 7000 | 0.2103          | 0.3288 |
| 0.8602        | 79.78 | 7500 | 0.2088          | 0.3238 |
| 0.8442        | 85.11 | 8000 | 0.2045          | 0.3266 |
| 0.8335        | 90.42 | 8500 | 0.2038          | 0.3241 |
| 0.8288        | 95.74 | 9000 | 0.2024          | 0.3280 |


### Framework versions

- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu102
- Datasets 1.18.3
- Tokenizers 0.11.0