hubert-base-ls960-finetuned-gtzan-efficient-label-smoothed

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: 2.2778
• Accuracy: 0.84

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: 8
• eval_batch_size: 8
• 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: 30
• label_smoothing_factor: 0.9

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2946	1.0	113	2.2960	0.42
2.2821	2.0	226	2.2858	0.63
2.2842	3.0	339	2.2848	0.68
2.2685	4.0	452	2.2836	0.71
2.2688	5.0	565	2.2838	0.71
2.2901	6.0	678	2.2804	0.75
2.2683	7.0	791	2.2831	0.7
2.2695	8.0	904	2.2833	0.75
2.268	9.0	1017	2.2793	0.8
2.2836	10.0	1130	2.2867	0.68
2.2704	11.0	1243	2.2811	0.78
2.2665	12.0	1356	2.2783	0.83
2.2663	13.0	1469	2.2782	0.85
2.2669	14.0	1582	2.2759	0.88
2.266	15.0	1695	2.2800	0.82
2.266	16.0	1808	2.2797	0.82
2.266	17.0	1921	2.2778	0.84

Framework versions

• Transformers 4.30.2
• Pytorch 2.1.0.dev20230627+cu121
• Datasets 2.13.1
• Tokenizers 0.13.3