Lifan-Z/hubert-base-ls960-finetuned-gtzan

hubert-base-ls960-v2-finetuned-gtzan

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

• Loss: 0.6560
• Accuracy: 0.86

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: 10
• eval_batch_size: 10
• 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: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.1778	1.0	90	2.1185	0.42
1.7477	2.0	180	1.6950	0.5
1.6626	3.0	270	1.4481	0.49
1.0488	4.0	360	1.2952	0.56
0.9819	5.0	450	1.0239	0.63
0.8553	6.0	540	0.8149	0.75
0.9188	7.0	630	0.9471	0.73
0.5563	8.0	720	0.7414	0.77
0.6793	9.0	810	0.7851	0.78
0.5282	10.0	900	0.6163	0.8
0.3895	11.0	990	0.6667	0.82
0.3037	12.0	1080	0.6157	0.84
0.1647	13.0	1170	0.6485	0.83
0.3331	14.0	1260	0.5609	0.86
0.1695	15.0	1350	0.6393	0.84
0.0968	16.0	1440	0.7537	0.83
0.1928	17.0	1530	0.7043	0.86
0.1281	18.0	1620	0.6077	0.89
0.0482	19.0	1710	0.7178	0.86
0.1215	20.0	1800	0.6560	0.86

Framework versions

• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1