m-a-p/MERT-v1-330M-finetuned-gtzan

This model is a fine-tuned version of m-a-p/MERT-v1-330M on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 0.4795
• Accuracy: 0.93

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: 16
• 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_ratio: 0.1
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.125	1.0	57	2.0967	0.46
1.1862	2.0	114	1.0739	0.73
0.7055	3.0	171	0.7744	0.81
0.445	4.0	228	0.5283	0.86
0.2788	5.0	285	0.4848	0.89
0.0948	6.0	342	0.5192	0.89
0.0709	7.0	399	0.4638	0.9
0.0818	8.0	456	0.4345	0.91
0.0835	9.0	513	0.4638	0.92
0.0063	10.0	570	0.4175	0.94
0.0052	11.0	627	0.4807	0.93
0.0055	12.0	684	0.4814	0.93
0.0235	13.0	741	0.4796	0.93
0.0043	14.0	798	0.4793	0.93
0.0044	15.0	855	0.4795	0.93

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1