whisper-base-ru

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

• Loss: 0.4816
• Model Preparation Time: 0.0058
• Wer: 33.1041

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: 5e-06
• 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: linear
• lr_scheduler_warmup_steps: 0.1
• num_epochs: 6
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Wer
0.7825	0.7752	300	0.7634	0.0058	46.2482
0.6010	1.5504	600	0.6149	0.0058	39.2065
0.4511	2.3256	900	0.5522	0.0058	37.9747
0.3867	3.1008	1200	0.5185	0.0058	35.1365
0.3590	3.8760	1500	0.4979	0.0058	34.2640
0.3290	4.6512	1800	0.4875	0.0058	33.8945
0.3170	5.4264	2100	0.4830	0.0058	33.5609
0.3134	6.0	2322	0.4816	0.0058	33.1041

Framework versions

• Transformers 5.13.1
• Pytorch 2.6.0+cu124
• Datasets 5.0.0
• Tokenizers 0.22.2