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: 0.6494
• Accuracy: 0.82

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• 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: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2235	1.0	57	2.1240	0.46
1.6742	2.0	114	1.5934	0.57
1.2803	3.0	171	1.3244	0.62
0.9666	4.0	228	1.0472	0.72
0.6942	5.0	285	0.8506	0.83
0.6772	6.0	342	0.7223	0.83
0.4997	7.0	399	0.7492	0.79
0.3602	8.0	456	0.7090	0.8
0.3209	9.0	513	0.6171	0.81
0.1864	10.0	570	0.5821	0.82
0.1642	11.0	627	0.5885	0.82
0.1225	12.0	684	0.7474	0.81
0.1189	13.0	741	0.6764	0.84
0.0773	14.0	798	0.6835	0.81
0.074	14.7467	840	0.6494	0.82

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1