hubert-large-ls960-ft-finetuned-gtzan

This model is a fine-tuned version of facebook/hubert-large-ls960-ft on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 0.7096
• Accuracy: 0.85

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: 8e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• 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
2.2623	1.0	56	2.2399	0.21
1.881	1.99	112	1.7105	0.41
1.5793	2.99	168	1.6203	0.46
1.3018	4.0	225	1.3824	0.52
1.0219	5.0	281	0.9899	0.66
0.9047	5.99	337	0.8812	0.74
0.8353	6.99	393	0.7629	0.78
0.659	8.0	450	0.9674	0.71
0.645	9.0	506	0.8953	0.74
0.6233	9.99	562	0.6638	0.8
0.4155	10.99	618	0.6323	0.81
0.2689	12.0	675	0.5423	0.83
0.3714	13.0	731	0.6770	0.83
0.0692	13.99	787	0.6260	0.83
0.0778	14.99	843	0.5801	0.85
0.187	16.0	900	0.6722	0.83
0.1469	17.0	956	0.7473	0.85
0.1052	17.92	1008	0.7096	0.85

Framework versions

• Transformers 4.30.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3