Garon16/whisper_small_ru_f

Whisper Small CV 18 - Garon

This model is a fine-tuned version of openai/whisper-small on the Common Voice 18.0 Ru dataset. It achieves the following results on the evaluation set:

• Loss: 0.2906
• Wer: 16.3383

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: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2603	1.0	826	0.2231	18.2503
0.1298	2.0	1652	0.2108	17.0453
0.0586	3.0	2478	0.2165	16.8375
0.0271	4.0	3304	0.2315	16.7760
0.0122	5.0	4130	0.2478	16.7864
0.0057	6.0	4956	0.2667	16.5670
0.0029	7.0	5782	0.2727	16.2594
0.0018	8.0	6608	0.2833	16.3743
0.0013	9.0	7434	0.2885	16.2594
0.0011	10.0	8260	0.2906	16.3383

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0