MMS-Wolof-1-hour-Alffa-dataset

This model is a fine-tuned version of facebook/mms-1b-all on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.2106
• Wer: 0.9858
• Cer: 0.1711

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: 8
• eval_batch_size: 16
• 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_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
2.0137	3.7383	200	0.6938	0.9988	0.1838
0.7405	7.4766	400	0.8238	1.0	0.1898
0.5635	11.2150	600	0.9400	1.0	0.1983
0.4271	14.9533	800	0.8553	0.9976	0.1909
0.3121	18.6916	1000	0.9862	0.9941	0.1846
0.2337	22.4299	1200	1.0075	0.9976	0.1855
0.1925	26.1682	1400	1.0512	0.9965	0.1830
0.1593	29.9065	1600	1.1660	0.9965	0.1918
0.1367	33.6449	1800	1.1406	0.9905	0.1829
0.1111	37.3832	2000	1.1527	0.9965	0.1842
0.0958	41.1215	2200	1.1590	0.9882	0.1790
0.0774	44.8598	2400	1.2424	0.9882	0.1802
0.0624	48.5981	2600	1.2106	0.9858	0.1711

Framework versions

• Transformers 4.45.1
• Pytorch 2.1.0+cu118
• Datasets 2.17.0
• Tokenizers 0.20.2