sitwala/whisper-large-v3-turbo-anv-zul

Whisper whisper-large-v3-turbo xho

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the dsfsi-anv/za-african-next-voices dataset. It achieves the following results on the evaluation set:

• Loss: 0.3618
• Wer: 19.6309

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: 64
• eval_batch_size: 64
• 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: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.3576	1.007	500	0.3938	29.3971
0.2536	2.014	1000	0.3147	22.4916
0.1773	3.021	1500	0.3101	21.7084
0.1419	4.028	2000	0.3081	20.6204
0.0905	5.035	2500	0.3142	20.8713
0.0611	6.042	3000	0.3231	20.3719
0.0617	7.049	3500	0.3326	20.4680
0.036	8.056	4000	0.3375	20.4235
0.0304	9.063	4500	0.3595	20.3766
0.0208	10.07	5000	0.3618	19.6309

Framework versions

• Transformers 4.52.0
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.4