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: 0.9570
• Accuracy: 0.86

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: 2
• 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.1586	1.0	112	2.0855	0.45
1.4771	2.0	225	1.3396	0.72
1.181	3.0	337	0.9735	0.76
0.8133	4.0	450	0.8692	0.76
0.5397	5.0	562	0.7118	0.81
0.3424	6.0	675	0.6237	0.81
0.2717	7.0	787	0.6551	0.83
0.2653	8.0	900	0.6707	0.83
0.0503	9.0	1012	0.7025	0.84
0.0168	10.0	1125	0.7643	0.87
0.1125	11.0	1237	0.8550	0.86
0.155	12.0	1350	0.9796	0.82
0.005	13.0	1462	0.9539	0.86
0.0038	14.0	1575	0.9206	0.86
0.0035	15.0	1687	0.8725	0.88
0.051	16.0	1800	0.9980	0.86
0.003	17.0	1912	0.9579	0.86
0.0025	18.0	2025	0.9735	0.86
0.0023	19.0	2137	0.9589	0.86
0.0022	19.91	2240	0.9570	0.86

Framework versions

• Transformers 4.29.2
• Pytorch 2.0.0
• Datasets 2.13.1
• Tokenizers 0.13.3