w2v2-bert-r-Wolof-5-hours-kallaama-dataset

This model is a fine-tuned version of facebook/w2v-bert-2.0 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.3781
• Wer: 0.5466
• Cer: 0.2727

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: 8
• seed: 42
• gradient_accumulation_steps: 4
• 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
2.299	5.9480	400	1.9903	0.8333	0.4075
1.721	11.8959	800	1.7083	0.7612	0.4171
1.3926	17.8439	1200	1.6886	0.6504	0.3240
1.1457	23.7918	1600	1.5706	0.6120	0.3168
0.9292	29.7398	2000	1.7518	0.5761	0.2857
0.7281	35.6877	2400	1.8207	0.5500	0.2746
0.5193	41.6357	2800	1.8834	0.5599	0.2794
0.3446	47.5836	3200	2.3781	0.5466	0.2727

Framework versions

• Transformers 4.44.1
• Pytorch 2.1.0+cu118
• Datasets 2.17.0
• Tokenizers 0.19.1