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:

• Accuracy: 0.88
• Loss: 0.5101

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 19
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss
2.1142	1.0	57	0.5	1.9842
1.5086	2.0	114	0.63	1.4646
1.1112	3.0	171	0.76	1.1176
1.0085	4.0	228	0.74	0.9412
0.7851	5.0	285	0.8	0.7978
0.6372	6.0	342	0.78	0.7533
0.5404	7.0	399	0.75	0.7206
0.4701	8.0	456	0.8	0.6551
0.4362	9.0	513	0.77	0.6712
0.3737	10.0	570	0.81	0.6202
0.321	11.0	627	0.78	0.6756
0.2533	12.0	684	0.84	0.5602
0.326	13.0	741	0.84	0.5706
0.1789	14.0	798	0.83	0.5736
0.1841	15.0	855	0.85	0.5379
0.2496	16.0	912	0.87	0.5518
0.2002	17.0	969	0.86	0.5220
0.1164	18.0	1026	0.86	0.5213
0.096	19.0	1083	0.88	0.5101

Framework versions

• Transformers 4.37.2
• Pytorch 2.3.0
• Datasets 2.19.0
• Tokenizers 0.15.1