RoBerta_pubHealth

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5020
• Accuracy: 0.795
• Auc: 0.883
• Precision: 0.908
• Recall: 0.738
• F1: 0.814
• F1-macro: 0.793
• F1-micro: 0.795
• F1-weighted: 0.798

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• 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
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Auc	Precision	Recall	F1	F1-macro	F1-micro	F1-weighted
0.5384	0.3960	100	0.4643	0.773	0.838	0.887	0.718	0.793	0.771	0.773	0.776
0.4741	0.7921	200	0.4453	0.779	0.843	0.93	0.688	0.791	0.778	0.779	0.781
0.43	1.1861	300	0.4406	0.791	0.857	0.909	0.73	0.809	0.789	0.791	0.794
0.4253	1.5822	400	0.4639	0.74	0.85	0.972	0.588	0.733	0.739	0.74	0.738
0.4126	1.9782	500	0.4164	0.784	0.869	0.933	0.694	0.796	0.783	0.784	0.786
0.3775	2.3723	600	0.4987	0.739	0.854	0.983	0.579	0.729	0.738	0.739	0.736
0.3617	2.7683	700	0.4202	0.786	0.875	0.915	0.715	0.802	0.785	0.786	0.789
0.3309	3.1624	800	0.4239	0.787	0.883	0.895	0.736	0.808	0.785	0.787	0.79
0.3069	3.5584	900	0.4535	0.805	0.887	0.889	0.776	0.829	0.802	0.805	0.808
0.3071	3.9545	1000	0.4527	0.801	0.89	0.919	0.738	0.819	0.8	0.801	0.804
0.2605	4.3485	1100	0.4896	0.783	0.879	0.887	0.736	0.805	0.781	0.783	0.786
0.2609	4.7446	1200	0.5020	0.795	0.883	0.908	0.738	0.814	0.793	0.795	0.798

Framework versions

• Transformers 4.53.0
• Pytorch 2.6.0+cu124
• Datasets 2.14.4
• Tokenizers 0.21.2