whisper-large-v3-ft-btb-cv-ca-cy

This model is a fine-tuned version of openai/whisper-large-v3 on the DewiBrynJones/banc-trawsgrifiadau-bangor-clean train main, DewiBrynJones/commonvoice_18_0_cy train+dev+other_with_excluded main, cymen-arfor/25awr train+dev main dataset. It achieves the following results on the evaluation set:

• Loss: 0.3666
• Wer: 0.2773

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.5072	0.4068	1000	0.5020	0.3567
0.3708	0.8137	2000	0.4260	0.3258
0.2599	1.2205	3000	0.3973	0.3003
0.2618	1.6273	4000	0.3783	0.2905
0.1846	2.0342	5000	0.3666	0.2773

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3