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: 0.9405
• Accuracy: 0.83

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• 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
2.1782	1.0	113	2.0403	0.46
1.487	2.0	226	1.3902	0.64
1.1634	3.0	339	1.0882	0.71
0.9874	4.0	452	0.9260	0.68
0.7754	5.0	565	0.7265	0.8
0.4512	6.0	678	0.6141	0.84
0.4947	7.0	791	0.8277	0.78
0.1896	8.0	904	0.7220	0.81
0.2142	9.0	1017	0.6393	0.85
0.0413	10.0	1130	0.8113	0.82
0.0105	11.0	1243	0.7368	0.82
0.1392	12.0	1356	0.8139	0.85
0.0051	13.0	1469	0.7893	0.86
0.0041	14.0	1582	0.8515	0.83
0.0041	15.0	1695	0.7707	0.85
0.0033	16.0	1808	0.8931	0.84
0.0772	17.0	1921	0.8411	0.86
0.0028	18.0	2034	0.8884	0.83
0.0025	19.0	2147	0.9094	0.84
0.0027	20.0	2260	0.9405	0.83

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0