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: 0.7345
• Accuracy: 0.87

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 6
• total_train_batch_size: 12
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.2
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2637	1.0	75	2.2059	0.34
1.8944	2.0	150	1.8194	0.41
1.5462	3.0	225	1.4462	0.6
1.27	4.0	300	1.1931	0.66
1.0759	5.0	375	0.9130	0.76
0.6731	6.0	450	0.8307	0.75
0.5021	7.0	525	0.6785	0.82
0.351	8.0	600	0.6946	0.8
0.259	9.0	675	0.5913	0.82
0.1789	10.0	750	0.6499	0.83
0.0655	11.0	825	0.5624	0.88
0.1194	12.0	900	0.6549	0.83
0.0874	13.0	975	0.6412	0.86
0.0142	14.0	1050	0.7119	0.86
0.0119	15.0	1125	0.7415	0.85
0.0093	16.0	1200	0.6833	0.87
0.0089	17.0	1275	0.7802	0.85
0.0142	18.0	1350	0.7611	0.85
0.0072	19.0	1425	0.7262	0.86
0.057	20.0	1500	0.7345	0.87

Framework versions

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