whisper-large-v3-ft-btb-ccv-cy

This model is a fine-tuned version of openai/whisper-large-v3 on the DewiBrynJones/banc-trawsgrifiadau-bangor-clean-with-ccv default dataset. It achieves the following results on the evaluation set:

• Loss: 0.4191
• Wer: 0.2911

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• 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
0.408	0.6311	1000	0.5	0.3734
0.2455	1.2622	2000	0.4340	0.3233
0.216	1.8933	3000	0.4010	0.3000
0.1308	2.5245	4000	0.4069	0.2964
0.0804	3.1556	5000	0.4191	0.2911

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1