whisper-large-v3-turbo-ft-cv-cy-train-all-plus-other-with-excluded

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the DewiBrynJones/commonvoice_18_0_cy train_all+other_with_excluded main dataset. It achieves the following results on the evaluation set:

• Loss: 0.3134
• Wer: 0.1746

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
0.2147	1.4144	1000	0.3066	0.2349
0.0989	2.8289	2000	0.2775	0.2072
0.0295	4.2433	3000	0.2935	0.1919
0.0109	5.6577	4000	0.3011	0.1828
0.0016	7.0721	5000	0.3134	0.1746

Framework versions

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