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: 0.3597
• Accuracy: 0.92

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: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5671	0.9956	112	0.5463	0.85
0.7083	2.0	225	0.6822	0.78
0.2257	2.9956	337	0.5415	0.85
0.028	4.0	450	0.5070	0.9
0.0526	4.9956	562	0.8882	0.82
0.0628	6.0	675	0.9979	0.79
0.0025	6.9956	787	0.5942	0.88
0.0005	8.0	900	0.6327	0.9
0.0005	8.9956	1012	0.4033	0.9
0.0009	10.0	1125	0.4190	0.88
0.0001	10.9956	1237	0.3672	0.93
0.0001	12.0	1350	0.3615	0.91
0.0001	12.9956	1462	0.3631	0.92
0.0001	14.0	1575	0.3597	0.92
0.0001	14.9956	1687	0.3604	0.92
0.0	16.0	1800	0.3589	0.92
0.0	16.9956	1912	0.3597	0.92
0.0434	18.0	2025	0.3590	0.92
0.0	18.9956	2137	0.3594	0.92
0.0	19.9111	2240	0.3597	0.92

Framework versions

• Transformers 4.41.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1