whisper-finetuned-v3_15e_augment

This model is a fine-tuned version of openai/whisper-large-v3-turbo on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0618
• Wer: 42.9388
• Cer: 25.8427

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• 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: 500
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.1455	1.0	1178	0.1225	66.9805	32.9972
0.0709	2.0	2356	0.0756	56.7249	28.5366
0.0434	3.0	3534	0.0650	52.0847	28.0039
0.0275	4.0	4712	0.0673	51.9502	27.1703
0.0153	5.0	5890	0.0571	48.6550	26.8105
0.0091	6.0	7068	0.0606	48.0498	26.7170
0.0063	7.0	8246	0.0622	45.9314	26.2351
0.002	8.0	9424	0.0609	44.7545	26.1131
0.0013	9.0	10602	0.0618	43.8467	25.9789
0.0007	9.9919	11770	0.0618	42.9388	25.8427

Framework versions

• Transformers 4.51.3
• Pytorch 2.7.0+cu126
• Datasets 3.5.1
• Tokenizers 0.21.1