distilhubert-finetuned-gtzan

This model is a fine-tuned version of ntu-spml/distilhubert on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 0.5114
• Accuracy: 0.88

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.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_ratio: 0.1
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.1662	1.0	113	2.0969	0.49
1.4471	2.0	226	1.4806	0.63
1.2335	3.0	339	1.1547	0.7
0.7599	4.0	452	0.8350	0.77
0.7346	5.0	565	0.7082	0.8
0.758	6.0	678	0.6305	0.75
0.4213	7.0	791	0.5270	0.86
0.1611	8.0	904	0.6318	0.83
0.3524	9.0	1017	0.5654	0.86
0.2389	10.0	1130	0.6017	0.83
0.0697	11.0	1243	0.5756	0.82
0.1679	12.0	1356	0.5597	0.86
0.0564	13.0	1469	0.7210	0.83
0.0394	14.0	1582	0.6780	0.85
0.0125	15.0	1695	0.7480	0.82
0.0147	16.0	1808	0.6366	0.83
0.1147	17.0	1921	0.6137	0.86
0.0083	18.0	2034	0.5979	0.85
0.0132	19.0	2147	0.6684	0.88
0.0064	20.0	2260	0.5114	0.88

Framework versions

• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.2