afrimbart_fr_Hausa

This model is a fine-tuned version of masakhane/afrimbart_fr_bam_news on the None dataset. It achieves the following results on the evaluation set:

• Bleu: 0.5099
• F1: 0.7516
• Wer: 0.3338
• Cer: 0.1229
• Meteor: 0.7261
• Loss: 7.9450

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: 0.0002
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Bleu	F1	Wer	Cer	Meteor	Validation Loss
8.0532	1.0	250	0.4830	0.7342	0.3570	0.1152	0.7008	7.9878
8.0259	2.0	500	0.5027	0.7481	0.3408	0.1311	0.7194	7.9523
8.0286	3.0	750	0.5099	0.7516	0.3338	0.1229	0.7261	7.9450

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 2.14.4
• Tokenizers 0.21.1