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: 0.6676
• Accuracy: 0.81

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: 3e-05
• train_batch_size: 6
• eval_batch_size: 6
• seed: 42
• gradient_accumulation_steps: 2
• 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: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2592	1.0	75	2.2017	0.35
1.8413	2.0	150	1.8071	0.45
1.5432	3.0	225	1.4808	0.65
1.2137	4.0	300	1.2621	0.7
1.1546	5.0	375	1.0581	0.77
0.9996	6.0	450	0.9858	0.75
0.7508	7.0	525	0.9087	0.78
0.6669	8.0	600	0.7710	0.81
0.6834	9.0	675	0.7663	0.8
0.4495	10.0	750	0.7184	0.79
0.3677	11.0	825	0.6589	0.81
0.3092	12.0	900	0.7223	0.8
0.1846	13.0	975	0.6665	0.82
0.1797	14.0	1050	0.6500	0.8
0.1695	15.0	1125	0.6549	0.81
0.1104	16.0	1200	0.6636	0.81
0.1192	17.0	1275	0.6722	0.81
0.1226	18.0	1350	0.6540	0.82
0.1218	19.0	1425	0.6646	0.79
0.067	20.0	1500	0.6676	0.81

Framework versions

• Transformers 4.33.0.dev0
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3