layoutlmv3-large-model2aa-visit-vs-progress

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: 0.3959
• Accuracy: 0.8680
• Macro Precision: 0.8680
• Macro Recall: 0.8680
• Macro F1: 0.8680
• Weighted F1: 0.8680
• Precision Visit Note Multiple Notes: 0.8644
• Recall Visit Note Multiple Notes: 0.8724
• F1 Visit Note Multiple Notes: 0.8684
• Precision Progress Follow Up Note: 0.8717
• Recall Progress Follow Up Note: 0.8636
• F1 Progress Follow Up Note: 0.8676

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	Accuracy	F1 Progress Follow Up Note	F1 Visit Note Multiple Notes	Validation Loss	Macro F1	Macro Precision	Macro Recall	Precision Progress Follow Up Note	Precision Visit Note Multiple Notes	Recall Progress Follow Up Note	Recall Visit Note Multiple Notes	Weighted F1
0.4912	0.3168	500	0.7062	0.6104	0.7641	0.5445	0.6873	0.7731	0.7066	0.9087	0.6374	0.4595	0.9537	0.6871
0.3608	0.6335	1000	0.8282	0.8418	0.8120	0.4213	0.8269	0.8378	0.8280	0.7812	0.8945	0.9126	0.7434	0.8269
0.3789	0.9503	1500	0.8485	0.8459	0.8509	0.3537	0.8484	0.8489	0.8485	0.8621	0.8357	0.8303	0.8667	0.8484
0.2658	1.2667	2000	0.8602	0.8596	0.8608	0.3743	0.8602	0.8602	0.8602	0.8648	0.8556	0.8544	0.8660	0.8602
0.3878	1.5835	2500	0.8641	0.8614	0.8667	0.3856	0.8641	0.8648	0.8641	0.8806	0.8489	0.8430	0.8852	0.8640
0.2924	1.9003	3000	0.8541	0.8499	0.8581	0.4044	0.8540	0.8554	0.8542	0.8769	0.8339	0.8246	0.8838	0.8540
0.3093	2.2167	3500	0.8680	0.8677	0.8683	0.3957	0.8680	0.8680	0.8680	0.8712	0.8649	0.8643	0.8717	0.8680
0.171	2.5334	4000	0.5077	0.8620	0.8632	0.8620	0.8619	0.8619	0.8418	0.8909	0.8657	0.8846	0.8331	0.8581
0.1688	2.8502	4500	0.5981	0.8623	0.8641	0.8624	0.8622	0.8622	0.8382	0.8974	0.8668	0.8900	0.8274	0.8576

Framework versions

• Transformers 4.57.6
• Pytorch 2.10.0+cu128
• Tokenizers 0.22.2