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: 0.5658
• Accuracy: 0.87
• F1: 0.87

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: 2
• eval_batch_size: 2
• seed: 2024
• 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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.8357	0.9956	56	0.6582	0.82	0.82
0.4742	1.9911	112	0.6527	0.81	0.81
0.3344	2.9867	168	0.9048	0.76	0.76
0.0659	4.0	225	0.6998	0.84	0.8400
0.0966	4.9956	281	0.6737	0.83	0.83
0.0026	5.9911	337	0.5133	0.89	0.89
0.0038	6.9867	393	0.5704	0.86	0.8600
0.0005	8.0	450	0.5722	0.86	0.8600
0.0003	8.9956	506	0.5632	0.87	0.87
0.0003	9.9556	560	0.5658	0.87	0.87

Framework versions

• Transformers 4.42.4
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1