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: 0.3539
• Accuracy: 0.91

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 18

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2281	1.0	112	2.1128	0.26
1.7082	2.0	225	1.6252	0.52
1.267	3.0	337	1.3100	0.54
1.1791	4.0	450	1.0496	0.71
1.1765	5.0	562	0.8928	0.74
0.5714	6.0	675	0.8298	0.77
0.4869	7.0	787	0.7145	0.79
0.4967	8.0	900	0.6990	0.82
0.8314	9.0	1012	0.5657	0.83
0.4633	10.0	1125	0.4589	0.89
0.5547	11.0	1237	0.4919	0.86
0.4827	12.0	1350	0.4069	0.92
0.324	13.0	1462	0.4634	0.87
0.5224	14.0	1575	0.4419	0.86
0.1873	15.0	1687	0.3988	0.89
0.2852	16.0	1800	0.3788	0.9
0.3169	17.0	1912	0.3526	0.89
0.4491	17.92	2016	0.3539	0.91

Framework versions

• Transformers 4.32.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3