senthilsdglakhsg/distilhubert-gtzan

distilhubert-finetuned-gtzan

This model is a fine-tuned version of ntu-spml/distilhubert on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 1.0795
• Accuracy: 0.83

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• 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
3.3659	1.0	113	1.5530	0.53
2.1453	2.0	226	1.1867	0.61
1.2708	3.0	339	0.8824	0.74
1.0719	4.0	452	0.8946	0.73
0.6139	5.0	565	0.8437	0.69
0.5381	6.0	678	0.7437	0.76
0.2904	7.0	791	0.8278	0.82
0.0359	8.0	904	1.1736	0.79
0.0341	9.0	1017	1.0795	0.83
0.0031	10.0	1130	1.1588	0.8
0.0020	11.0	1243	1.1727	0.81
0.0016	12.0	1356	1.2418	0.83
0.0013	13.0	1469	1.2516	0.82
0.0015	14.0	1582	1.3328	0.82
0.0010	15.0	1695	1.3100	0.82
0.0009	16.0	1808	1.3831	0.8
0.0009	17.0	1921	1.2942	0.82
0.0007	18.0	2034	1.3562	0.81
0.0007	19.0	2147	1.3046	0.81
0.0007	20.0	2260	1.3027	0.82

Framework versions

• Transformers 5.16.1
• Pytorch 2.11.0+cu128
• Datasets 5.0.1
• Tokenizers 0.23.1