layoutlmv3-large-model2b-radiology-lab-operative

This model is a fine-tuned version of microsoft/layoutlmv3-large on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1427
• Accuracy: 0.9689
• Macro Precision: 0.9680
• Macro Recall: 0.9696
• Macro F1: 0.9688
• Weighted F1: 0.9689
• Precision Radiology Report: 0.9752
• Recall Radiology Report: 0.9689
• F1 Radiology Report: 0.9720
• Precision Lab Report: 0.9589
• Recall Lab Report: 0.9668
• F1 Lab Report: 0.9628
• Precision Operative Report: 0.9699
• Recall Operative Report: 0.9732
• F1 Operative Report: 0.9715

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: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAFACTOR and the args are: No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Macro Precision	Macro Recall	Macro F1	Weighted F1	Precision Radiology Report	Recall Radiology Report	F1 Radiology Report	Precision Lab Report	Recall Lab Report	F1 Lab Report	Precision Operative Report	Recall Operative Report	F1 Operative Report
0.2022	0.4851	500	0.1871	0.9618	0.9581	0.9599	0.9589	0.9619	0.975	0.9614	0.9682	0.9465	0.9685	0.9574	0.9529	0.9497	0.9513
0.0799	0.9703	1000	0.1950	0.9618	0.9538	0.9639	0.9586	0.9619	0.9813	0.9550	0.9680	0.9481	0.9701	0.9590	0.9320	0.9664	0.9489
0.0367	1.4551	1500	0.2264	0.9635	0.9604	0.9645	0.9624	0.9635	0.9792	0.9603	0.9697	0.9418	0.9668	0.9542	0.96	0.9664	0.9632
0.0602	1.9402	2000	0.1863	0.9640	0.9620	0.9634	0.9626	0.9641	0.9751	0.9636	0.9693	0.9480	0.9668	0.9573	0.9630	0.9597	0.9613
0.0604	2.4250	2500	0.1429	0.9689	0.9680	0.9696	0.9688	0.9689	0.9752	0.9689	0.9720	0.9589	0.9668	0.9628	0.9699	0.9732	0.9715
0.0447	2.9101	3000	0.1817	0.9673	0.9673	0.9673	0.9672	0.9673	0.9772	0.9636	0.9703	0.9484	0.9751	0.9616	0.9762	0.9631	0.9696

Framework versions

• Transformers 4.57.6
• Pytorch 2.10.0+cu128
• Tokenizers 0.22.2