hubert-model-v2

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: 0.9612
• Accuracy: 0.82
• Precision: 0.8392
• Recall: 0.82
• F1: 0.8224

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: 4
• eval_batch_size: 4
• 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: 12
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
2.2086	1.0	200	2.0834	0.265	0.1987	0.265	0.1723
1.8657	2.0	400	1.7149	0.385	0.3228	0.385	0.3226
1.5242	3.0	600	1.4365	0.49	0.4463	0.49	0.4280
1.2542	4.0	800	1.2907	0.57	0.6029	0.57	0.5477
1.0119	5.0	1000	1.0524	0.69	0.7249	0.69	0.6745
0.8835	6.0	1200	1.1677	0.69	0.7156	0.69	0.6718
0.7271	7.0	1400	0.9158	0.745	0.7569	0.745	0.7309
0.5453	8.0	1600	0.7592	0.82	0.8296	0.82	0.8185
0.4503	9.0	1800	0.9936	0.775	0.8138	0.775	0.7799
0.3625	10.0	2000	0.9192	0.815	0.8258	0.815	0.8146
0.2773	11.0	2200	0.9768	0.82	0.8362	0.82	0.8201
0.2309	12.0	2400	0.9612	0.82	0.8392	0.82	0.8224

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0