whisper-small-ru-v6la

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

• Loss: 0.2438
• Wer: 17.6318

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 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: 200
• training_steps: 900
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2409	0.6098	100	0.2788	19.9434
0.1654	1.2195	200	0.2431	18.3040
0.1613	1.8293	300	0.2271	17.7851
0.0961	2.4390	400	0.2299	17.4195
0.0832	3.0488	500	0.2244	17.3252
0.0572	3.6585	600	0.2363	17.3370
0.0367	4.2683	700	0.2379	17.5492
0.0357	4.8780	800	0.2413	17.6908
0.0299	5.4878	900	0.2438	17.6318

Framework versions

• Transformers 4.49.0
• Pytorch 2.1.0+cu118
• Datasets 3.3.1
• Tokenizers 0.21.0