Hcask/distilhubert

hubert-base-ls960-finetuned-gtzan

This model is a fine-tuned version of facebook/hubert-base-ls960 on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 1.1234
• Accuracy: 0.7391

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.4277	1.0	25	1.5627	0.4783
1.4946	2.0	50	1.4727	0.5217
1.051	3.0	75	1.3207	0.6087
1.0897	4.0	100	1.3614	0.6522
1.1461	5.0	125	1.3143	0.5652
0.6919	6.0	150	1.1131	0.6087
0.7273	7.0	175	1.4138	0.6522
0.5955	8.0	200	1.2106	0.6957
0.4823	9.0	225	1.1681	0.6087
0.5178	10.0	250	1.1616	0.6522
0.4635	11.0	275	0.9685	0.7826
0.4622	12.0	300	0.9625	0.7826
0.3048	13.0	325	1.0364	0.7391
0.1576	14.0	350	1.0571	0.7391
0.1876	15.0	375	1.1234	0.7391

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1