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: 0.403864
• Accuracy: 0.88

Model description

ast-finetuned-audioset-10-10-0.4593

Intended uses & limitations

audio-classification

Training and evaluation data

marsyas/gtzan

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 0.1
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.7235	1.0	113	0.6009	0.85
0.3576	2.0	226	0.5851	0.81
0.0886	3.0	339	0.5113	0.85
0.0971	4.0	452	0.3983	0.86
0.0023	5.0	565	0.5785	0.87
0.0078	6.0	678	0.4070	0.88
0.0018	7.0	791	0.4260	0.88
0.0005	8.0	904	0.6328	0.85
0.0002	9.0	1017	0.4039	0.88
0.0002	10.0	1130	0.3952	0.87
0.0002	11.0	1243	0.3947	0.87
0.0001	12.0	1356	0.3944	0.87
0.0001	13.0	1469	0.3927	0.87
0.0001	14.0	1582	0.3930	0.87
0.0002	15.0	1695	0.3931	0.87

Framework versions

• Transformers 5.14.0.dev0
• Pytorch 2.11.0+cu128
• Datasets 3.6.0
• Tokenizers 0.22.2