DSU-ilabAfrica/whisper-swahili-large-v0.1

Whisper Swahili small v0.1

This model is a fine-tuned version of openai/whisper-small on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.8625
• Wer: 50.6499

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• 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
• lr_scheduler_warmup_steps: 25
• training_steps: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.3154	0.0333	25	1.1152	55.6473
0.7624	0.0667	50	0.9697	49.3212
0.7082	0.1	75	0.8964	51.6857
0.6339	0.1333	100	0.8625	50.6499

Framework versions

• Transformers 4.48.0
• Pytorch 2.6.0+cu126
• Datasets 3.5.0
• Tokenizers 0.21.4