hubert-base-ls960-2clsfinetuned-bmd-V1-epo10-20250201_141430-LOSO-section-out4

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

• Loss: 0.4886
• Accuracy: 0.7826
• F1: 0.7826
• Precision: 0.7856
• Recall: 0.7826

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: 1968
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• 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: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
No log	1.0	6	0.6795	0.5652	0.5145	0.6430	0.5652
0.6429	2.0	12	0.6508	0.6957	0.6922	0.7133	0.6957
0.6429	3.0	18	0.5901	0.7391	0.7288	0.7686	0.7391
0.5306	4.0	24	0.6976	0.6522	0.6441	0.6783	0.6522
0.4812	5.0	30	0.6430	0.6957	0.6590	0.8078	0.6957
0.4812	6.0	36	0.6919	0.6522	0.5963	0.7913	0.6522
0.4336	7.0	42	0.4892	0.7826	0.7826	0.7856	0.7826
0.4336	8.0	48	0.4931	0.7826	0.7826	0.7856	0.7826
0.3751	8.3810	50	0.4886	0.7826	0.7826	0.7856	0.7826

Framework versions

• Transformers 4.48.2
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0