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: 0.9983
• 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: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.0454	1.0	113	1.9653	0.56
1.3054	2.0	226	1.3132	0.7
0.9479	3.0	339	0.9596	0.75
0.7291	4.0	452	0.8110	0.75
0.6033	5.0	565	0.7330	0.81
0.2973	6.0	678	0.7070	0.79
0.3574	7.0	791	0.6908	0.83
0.2078	8.0	904	0.7105	0.83
0.1569	9.0	1017	0.7204	0.83
0.0812	10.0	1130	0.7471	0.84
0.0451	11.0	1243	0.8439	0.85
0.0148	12.0	1356	0.9538	0.83
0.0096	13.0	1469	0.9364	0.84
0.0084	14.0	1582	0.9808	0.83
0.0084	15.0	1695	0.9983	0.83

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.0.1
• Datasets 2.14.5
• Tokenizers 0.15.0