File size: 2,603 Bytes
aacca91
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
799d7fa
aacca91
 
 
 
 
 
 
 
 
799d7fa
 
aacca91
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
799d7fa
aacca91
 
f2157ca
799d7fa
aacca91
 
 
799d7fa
aacca91
 
 
f2157ca
 
799d7fa
 
 
 
 
 
 
 
 
 
 
 
 
 
aacca91
 
 
 
f2157ca
aacca91
2b18270
aacca91
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
90
91
92
---
license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
- generated_from_trainer
datasets:
- common_voice
metrics:
- wer
model-index:
- name: wav2vec2-large-xls-r-300m-euskera-colab
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: common_voice
      type: common_voice
      config: eu
      split: test
      args: eu
    metrics:
    - name: Wer
      type: wer
      value: 0.28292759459247446
---

<!-- 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. -->

# wav2vec2-large-xls-r-300m-euskera-colab

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: 0.2281
- Wer: 0.2829

## 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: 0.0003
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 6

### Training results

| Training Loss | Epoch | Step | Validation Loss | Wer    |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| 3.6474        | 0.43  | 400  | 0.8974          | 0.9025 |
| 0.4849        | 0.85  | 800  | 0.4653          | 0.6314 |
| 0.2924        | 1.28  | 1200 | 0.3726          | 0.5294 |
| 0.2392        | 1.7   | 1600 | 0.3203          | 0.4461 |
| 0.1957        | 2.13  | 2000 | 0.2932          | 0.4053 |
| 0.1592        | 2.56  | 2400 | 0.2767          | 0.3760 |
| 0.1442        | 2.98  | 2800 | 0.2605          | 0.3635 |
| 0.1166        | 3.41  | 3200 | 0.2662          | 0.3415 |
| 0.1064        | 3.84  | 3600 | 0.2576          | 0.3409 |
| 0.0906        | 4.26  | 4000 | 0.2567          | 0.3234 |
| 0.0818        | 4.69  | 4400 | 0.2472          | 0.3063 |
| 0.0701        | 5.11  | 4800 | 0.2440          | 0.2951 |
| 0.0595        | 5.54  | 5200 | 0.2321          | 0.2810 |
| 0.0566        | 5.97  | 5600 | 0.2281          | 0.2829 |


### Framework versions

- Transformers 4.32.1
- Pytorch 2.0.1+cu118
- Datasets 2.14.4
- Tokenizers 0.13.3