Lingua-Connect/whisper-small-sw-bible

Whisper_Small_swahili_bible

This model is a fine-tuned version of openai/whisper-small on the OpenBible_Swahili_book_split dataset. It achieves the following results on the evaluation set:

• Loss: 0.2543
• Wer Ortho: 19.1379
• Wer: 9.7620

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• 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: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.1658	1.0	1423	0.2543	19.1379	9.7620

Framework versions

• Transformers 4.50.2
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1