File size: 3,405 Bytes
8907871
 
29f6e03
 
 
 
a7cab3d
 
 
 
29f6e03
 
a7cab3d
 
 
 
 
 
 
 
 
 
 
 
 
 
8907871
 
29f6e03
 
8907871
29f6e03
 
8907871
a7cab3d
 
 
 
 
8907871
29f6e03
8907871
29f6e03
8907871
29f6e03
8907871
29f6e03
8907871
29f6e03
8907871
29f6e03
8907871
29f6e03
8907871
29f6e03
8907871
29f6e03
 
 
 
 
 
 
 
 
 
 
8907871
a7cab3d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
29f6e03
8907871
29f6e03
 
 
 
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
93
94
95
96
97
98
99
100
---
library_name: transformers
license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
- generated_from_trainer
datasets:
- fleurs
metrics:
- wer
model-index:
- name: wav2vec2-xls-r-Wolof-10-hours-Google-Fleurs-dataset
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: fleurs
      type: fleurs
      config: wo_sn
      split: None
      args: wo_sn
    metrics:
    - name: Wer
      type: wer
      value: 0.442296823782073
---

<!-- 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. -->

[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/asr-africa-research-team/ASR%20Africa/runs/1lxkt8t0)
# wav2vec2-xls-r-Wolof-10-hours-Google-Fleurs-dataset

This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the fleurs dataset.
It achieves the following results on the evaluation set:
- Loss: 1.2082
- Wer: 0.4423
- Cer: 0.1524

## 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: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 50

### Training results

| Training Loss | Epoch   | Step | Validation Loss | Wer    | Cer    |
|:-------------:|:-------:|:----:|:---------------:|:------:|:------:|
| 6.6995        | 2.6144  | 200  | 3.0428          | 1.0    | 1.0    |
| 3.0391        | 5.2288  | 400  | 3.0117          | 1.0    | 1.0    |
| 3.0035        | 7.8431  | 600  | 2.9794          | 1.0    | 1.0    |
| 2.0946        | 10.4575 | 800  | 0.9827          | 0.7357 | 0.2560 |
| 0.8407        | 13.0719 | 1000 | 0.7398          | 0.5189 | 0.1848 |
| 0.5774        | 15.6863 | 1200 | 0.7214          | 0.4926 | 0.1745 |
| 0.4229        | 18.3007 | 1400 | 0.6996          | 0.4852 | 0.1707 |
| 0.3332        | 20.9150 | 1600 | 0.7950          | 0.4878 | 0.1708 |
| 0.2488        | 23.5294 | 1800 | 0.8972          | 0.4645 | 0.1624 |
| 0.2043        | 26.1438 | 2000 | 0.9122          | 0.4576 | 0.1609 |
| 0.1699        | 28.7582 | 2200 | 1.0064          | 0.4777 | 0.1672 |
| 0.1472        | 31.3725 | 2400 | 1.0141          | 0.4554 | 0.1581 |
| 0.1251        | 33.9869 | 2600 | 1.0362          | 0.4553 | 0.1580 |
| 0.1152        | 36.6013 | 2800 | 1.1312          | 0.4490 | 0.1554 |
| 0.0986        | 39.2157 | 3000 | 1.1552          | 0.4499 | 0.1555 |
| 0.0905        | 41.8301 | 3200 | 1.1811          | 0.4463 | 0.1547 |
| 0.0879        | 44.4444 | 3400 | 1.1849          | 0.4513 | 0.1551 |
| 0.0793        | 47.0588 | 3600 | 1.2074          | 0.4422 | 0.1527 |
| 0.0802        | 49.6732 | 3800 | 1.2082          | 0.4423 | 0.1524 |


### Framework versions

- Transformers 4.44.2
- Pytorch 2.1.0+cu118
- Datasets 2.17.0
- Tokenizers 0.19.1