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: 1.0908
• 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: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.1531	1.0	113	2.1667	0.41
1.6622	2.0	226	1.6138	0.63
1.3112	3.0	339	1.2047	0.7
0.9374	4.0	452	0.9595	0.73
0.5475	5.0	565	0.7239	0.82
0.4845	6.0	678	0.7406	0.75
0.2489	7.0	791	0.6838	0.78
0.3272	8.0	904	0.8447	0.79
0.2244	9.0	1017	0.7184	0.81
0.0353	10.0	1130	0.8800	0.79
0.0201	11.0	1243	0.8800	0.83
0.0079	12.0	1356	0.8207	0.83
0.004	13.0	1469	0.9218	0.82
0.003	14.0	1582	1.0004	0.83
0.0024	15.0	1695	0.9446	0.84
0.0021	16.0	1808	0.9802	0.85
0.0018	17.0	1921	0.9766	0.84
0.0017	18.0	2034	1.0597	0.84
0.0014	19.0	2147	0.9541	0.84
0.0012	20.0	2260	1.0408	0.84
0.0011	21.0	2373	1.0364	0.84
0.001	22.0	2486	1.0993	0.84
0.001	23.0	2599	1.0620	0.84
0.0009	24.0	2712	1.0193	0.83
0.0009	25.0	2825	1.0164	0.83
0.0009	26.0	2938	1.0293	0.84
0.0008	27.0	3051	1.0478	0.84
0.0008	28.0	3164	1.0727	0.84
0.0008	29.0	3277	1.0773	0.84
0.0008	30.0	3390	1.0908	0.84

Framework versions

• Transformers 4.30.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.12.0
• Tokenizers 0.13.3