whisper-large-v3-basque

This model is a fine-tuned version of openai/whisper-large-v3 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1504
• Wer: 12.5989

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: 2e-06
• train_batch_size: 256
• eval_batch_size: 32
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• total_train_batch_size: 512
• total_eval_batch_size: 64
• 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
0.1457	0.66	500	0.2250	19.0026
0.0999	1.32	1000	0.1848	16.3283
0.0863	1.98	1500	0.1650	14.2673
0.0728	2.64	2000	0.1591	13.6110
0.063	3.3	2500	0.1547	13.5374
0.0601	3.96	3000	0.1498	12.8872
0.0549	4.62	3500	0.1500	12.4885
0.0499	5.28	4000	0.1504	12.5682
0.0487	5.94	4500	0.1498	12.7216
0.0477	6.61	5000	0.1504	12.5989

Framework versions

• Transformers 4.38.0
• Pytorch 2.1.1+cu121
• Datasets 2.8.0
• Tokenizers 0.15.2