ast-finetuned-audioset-10-10-0.4593-finetuned-gtzan

This model is a fine-tuned version of MIT/ast-finetuned-audioset-10-10-0.4593 on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 0.3715
• Accuracy: 0.93
• Precision: 0.9386
• Recall: 0.93
• F1: 0.9311

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: 16
• eval_batch_size: 16
• 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	Precision	Recall	F1
0.8552	1.0	57	0.5962	0.83	0.8693	0.83	0.8207
0.448	2.0	114	0.5167	0.85	0.8736	0.85	0.8534
0.1634	3.0	171	0.5433	0.86	0.8780	0.86	0.8570
0.1673	4.0	228	0.4743	0.88	0.8836	0.88	0.8769
0.0065	5.0	285	0.4956	0.91	0.9212	0.91	0.9060
0.0279	6.0	342	0.5635	0.89	0.8971	0.89	0.8879
0.104	7.0	399	0.6799	0.86	0.8832	0.86	0.8564
0.001	8.0	456	0.4927	0.91	0.9246	0.91	0.9109
0.0002	9.0	513	0.3899	0.92	0.9245	0.92	0.9187
0.0002	10.0	570	0.3715	0.93	0.9386	0.93	0.9311
0.0002	11.0	627	0.4695	0.92	0.9245	0.92	0.9180
0.0001	12.0	684	0.4150	0.93	0.9370	0.93	0.9291
0.0468	13.0	741	0.4483	0.92	0.9294	0.92	0.9182
0.0001	14.0	798	0.3852	0.93	0.9334	0.93	0.9288

Framework versions

• Transformers 4.42.3
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1