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: 0.6087
• Accuracy: 0.88

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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.9526	1.0	112	1.8797	0.74
0.9704	2.0	225	1.0561	0.7
0.7957	3.0	337	0.7362	0.77
0.4428	4.0	450	0.7820	0.8
0.1422	5.0	562	0.6142	0.84
0.3502	6.0	675	0.9189	0.82
0.01	7.0	787	0.7735	0.83
0.0068	8.0	900	1.0699	0.81
0.1751	9.0	1012	0.5360	0.88
0.0045	10.0	1125	0.5377	0.89
0.154	11.0	1237	0.6542	0.86
0.0025	12.0	1350	0.6206	0.89
0.0022	13.0	1462	0.6118	0.88
0.0021	14.0	1575	0.5961	0.89
0.0018	15.0	1687	0.5958	0.88
0.0017	16.0	1800	0.6062	0.88
0.0017	17.0	1912	0.6005	0.88
0.0015	18.0	2025	0.6052	0.88
0.0014	19.0	2137	0.6114	0.88
0.0015	19.91	2240	0.6087	0.88

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu118
• Datasets 2.16.1
• Tokenizers 0.15.0