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: 0.9432
• Accuracy: 0.84

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
2.1412	1.0	113	2.0696	0.55
1.433	2.0	226	1.4669	0.6
1.1466	3.0	339	1.0781	0.7
0.689	4.0	452	0.8115	0.78
0.5486	5.0	565	0.7290	0.79
0.3461	6.0	678	0.6226	0.81
0.3023	7.0	791	0.5584	0.8
0.1167	8.0	904	0.6853	0.81
0.0821	9.0	1017	0.6419	0.83
0.0398	10.0	1130	0.7172	0.83
0.0149	11.0	1243	0.7830	0.82
0.0081	12.0	1356	0.8100	0.83
0.0062	13.0	1469	0.8566	0.85
0.0049	14.0	1582	0.9472	0.83
0.0039	15.0	1695	0.9194	0.83
0.0036	16.0	1808	0.9292	0.84
0.0034	17.0	1921	0.9399	0.83
0.003	18.0	2034	0.9606	0.83
0.0028	19.0	2147	0.9453	0.84
0.0028	20.0	2260	0.9432	0.84

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2