YashwanthReddyL/gtzan_ast

AST-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: 0.4014
• 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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• 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: 100
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.7737	1.0	113	0.6396	0.85
0.3255	2.0	226	0.5410	0.85
0.1630	3.0	339	0.3864	0.9
0.1103	4.0	452	0.4014	0.86
0.0052	5.0	565	0.3631	0.91
0.0067	6.0	678	0.4079	0.88
0.0031	7.0	791	0.3714	0.88
0.0007	8.0	904	0.3700	0.92
0.0005	9.0	1017	0.3786	0.91
0.0004	10.0	1130	0.4100	0.91
0.0003	11.0	1243	0.4028	0.91
0.0003	12.0	1356	0.3877	0.91
0.0003	13.0	1469	0.3884	0.92
0.0003	14.0	1582	0.4290	0.91
0.0003	15.0	1695	0.4110	0.91
0.0002	16.0	1808	0.4178	0.91
0.0002	17.0	1921	0.3992	0.92
0.0002	18.0	2034	0.4116	0.91
0.0002	19.0	2147	0.4023	0.92
0.0002	20.0	2260	0.4014	0.92

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 2.18.0
• Tokenizers 0.22.2