whisper-small-shona

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

• Loss: 0.5078
• Wer: 0.4034

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: 32
• 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: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.0079	1.0753	500	0.4404	0.4562
0.3609	2.1505	1000	0.3641	0.3821
0.2676	3.2258	1500	0.3509	0.3837
0.2073	4.3011	2000	0.3607	0.3595
0.1587	5.3763	2500	0.3773	0.3688
0.1163	6.4516	3000	0.4026	0.3778
0.0803	7.5269	3500	0.4370	0.4012
0.0542	8.6022	4000	0.4641	0.4035
0.0366	9.6774	4500	0.4942	0.4008
0.0259	10.7527	5000	0.5078	0.4034

Framework versions

• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 2.14.4
• Tokenizers 0.21.2