ast-genre-classifier

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

• Loss: 1.6239
• Accuracy: 0.8725
• F1: 0.8716

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 16
• 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: cosine
• lr_scheduler_warmup_steps: 200
• num_epochs: 15
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
3.1456	1.0	113	3.0056	0.5787	0.5602
2.3584	2.0	226	2.4650	0.645	0.6353
2.0640	3.0	339	2.1529	0.7525	0.7514
1.9139	4.0	452	2.1192	0.7438	0.7280
1.8101	5.0	565	1.8596	0.815	0.8156
1.6885	6.0	678	1.9938	0.7725	0.7702
1.5576	7.0	791	1.8289	0.8237	0.8188
1.5289	8.0	904	1.7069	0.855	0.8554
1.3774	9.0	1017	1.7971	0.8237	0.8224
1.3125	10.0	1130	1.5755	0.8725	0.8722
1.2516	11.0	1243	1.7026	0.85	0.8480
1.2515	12.0	1356	1.6192	0.8575	0.8569
1.2180	13.0	1469	1.5849	0.8675	0.8669
1.2214	14.0	1582	1.5226	0.89	0.8895
1.1390	15.0	1695	1.6239	0.8725	0.8716

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.8.3
• Tokenizers 0.22.2