whisper-medium.en-finetuned-gtzan

This model is a fine-tuned version of openai/whisper-medium.en on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 0.2885
• Accuracy: 0.95

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 16
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.7711	1.0	112	1.6556	0.52
0.5477	2.0	225	0.4738	0.85
0.535	3.0	337	0.3137	0.92
0.231	4.0	450	0.3613	0.9
0.1923	5.0	562	0.2885	0.95
0.0584	6.0	675	0.6531	0.86
0.1783	7.0	787	0.5717	0.9
0.0022	8.0	900	0.4205	0.91
0.1032	9.0	1012	0.4984	0.91
0.0011	10.0	1125	0.3778	0.94
0.0104	11.0	1237	0.3709	0.94
0.0011	12.0	1350	0.4564	0.92
0.0009	13.0	1462	0.3796	0.94
0.0008	14.0	1575	0.3880	0.94
0.0008	15.0	1687	0.3930	0.94
0.0008	15.93	1792	0.3955	0.94

Framework versions

• Transformers 4.37.0.dev0
• Pytorch 2.1.2+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0