transfer-learning-finetuned-gtzan-model

This model is a fine-tuned version of sanchit-gandhi/distilhubert-finetuned-gtzan on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.3704
• Accuracy: 0.5565
• F1: 0.5318

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: 2e-05
• train_batch_size: 32
• 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: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.05

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.9103	0.4735	500	1.9692	0.2823	0.2697
1.6753	0.9470	1000	1.7129	0.3992	0.3692
1.4613	1.4205	1500	1.6354	0.4240	0.3987
1.4226	1.8939	2000	1.5426	0.4504	0.4267
1.3909	2.3674	2500	1.5176	0.4702	0.4423
1.3919	2.8409	3000	1.4944	0.4855	0.4584
1.2577	3.3144	3500	1.4466	0.4955	0.4710
1.2455	3.7879	4000	1.4501	0.5050	0.4833
1.1261	4.2614	4500	1.4460	0.4945	0.4757
1.1117	4.7348	5000	1.4453	0.5087	0.4812
1.0591	5.2083	5500	1.4648	0.5032	0.4738
1.079	5.6818	6000	1.3912	0.5288	0.4998
1.0071	6.1553	6500	1.3956	0.5285	0.5012
0.976	6.6288	7000	1.4003	0.5309	0.5049
0.9517	7.1023	7500	1.4035	0.5343	0.5079
0.9873	7.5758	8000	1.3696	0.5383	0.5140
0.8671	8.0492	8500	1.3814	0.5446	0.5217
0.8546	8.5227	9000	1.3788	0.5446	0.5173
0.9945	8.9962	9500	1.3796	0.5454	0.5229
0.857	9.4697	10000	1.3667	0.5530	0.5293
0.8703	9.9432	10500	1.3439	0.5536	0.5292
0.8018	10.4167	11000	1.3654	0.5554	0.5296
0.8333	10.8902	11500	1.3696	0.5491	0.5235
0.8282	11.3636	12000	1.3825	0.5512	0.5245
0.7193	11.8371	12500	1.3758	0.5554	0.5296
0.7366	12.3106	13000	1.3582	0.5594	0.5326
0.792	12.7841	13500	1.3829	0.5512	0.5262
0.7758	13.2576	14000	1.3684	0.5588	0.5340
0.8631	13.7311	14500	1.3663	0.5580	0.5325
0.8261	14.2045	15000	1.3752	0.5562	0.5310
0.7624	14.6780	15500	1.3704	0.5565	0.5318

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu126
• Datasets 4.4.2
• Tokenizers 0.22.1