whisper-lora-630

This model is a fine-tuned version of openai/whisper-large-v3 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.4043
• Wer: 0.5039
• Cer: 0.2762

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: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• 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: 50
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Cer	Validation Loss	Wer
1.4297	0.3361	10	0.2634	1.4343	0.4935
1.4671	0.6723	20	0.2776	1.4356	0.5013
1.5794	1.0336	30	0.2661	1.4332	0.4909
1.4176	1.3361	40	1.4319	0.5013	0.2670
1.5047	1.6723	50	1.4263	0.5039	0.2762
1.3984	2.0	60	1.4191	0.4961	0.2670
1.5092	2.3361	70	1.4130	0.4961	0.2675
1.4347	2.6723	80	1.4043	0.5039	0.2762

Framework versions

• PEFT 0.16.0
• Transformers 4.53.2
• Pytorch 2.7.1+cu118
• Datasets 3.1.0
• Tokenizers 0.21.2