whisper-large-v2-LORA-uk

This model is a fine-tuned version of openai/whisper-large-v2 on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2644
• Wer: 53.6960
• Cer: 23.9099

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: 0.001
• train_batch_size: 8
• eval_batch_size: 8
• 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: 500
• training_steps: 6000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.4082	0.4034	1000	0.3692	62.5451	26.5080
0.3111	0.8068	2000	0.3501	61.2485	30.0473
0.235	1.2102	3000	0.3057	58.1339	26.9828
0.2138	1.6136	4000	0.2875	45.6757	18.5387
0.0866	2.0169	5000	0.2660	55.1798	24.2983
0.0934	2.4203	6000	0.2644	53.6960	23.9099

Framework versions

• PEFT 0.15.2.dev0
• Transformers 4.52.0.dev0
• Pytorch 2.5.1+cu124
• Datasets 3.5.0
• Tokenizers 0.21.0