lmv2-g-rai-aRx-refill-230427

This model is a fine-tuned version of microsoft/layoutlmv2-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0176
• Dob Key Precision: 0.6846
• Dob Key Recall: 0.7007
• Dob Key F1: 0.6926
• Dob Key Number: 598
• Dob Value Precision: 0.9934
• Dob Value Recall: 0.9983
• Dob Value F1: 0.9958
• Dob Value Number: 599
• Patient Name Key Precision: 0.6852
• Patient Name Key Recall: 0.7089
• Patient Name Key F1: 0.6968
• Patient Name Key Number: 608
• Patient Name Value Precision: 0.9581
• Patient Name Value Recall: 0.9738
• Patient Name Value F1: 0.9659
• Patient Name Value Number: 611
• Provider Name Key Precision: 0.7875
• Provider Name Key Recall: 0.7903
• Provider Name Key F1: 0.7889
• Provider Name Key Number: 558
• Provider Name Value Precision: 0.9786
• Provider Name Value Recall: 0.9751
• Provider Name Value F1: 0.9768
• Provider Name Value Number: 562
• Overall Precision: 0.8460
• Overall Recall: 0.8575
• Overall F1: 0.8517
• Overall Accuracy: 0.9936

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Dob Key Precision	Dob Key Recall	Dob Key F1	Dob Key Number	Dob Value Precision	Dob Value Recall	Dob Value F1	Dob Value Number	Patient Name Key Precision	Patient Name Key Recall	Patient Name Key F1	Patient Name Key Number	Patient Name Value Precision	Patient Name Value Recall	Patient Name Value F1	Patient Name Value Number	Provider Name Key Precision	Provider Name Key Recall	Provider Name Key F1	Provider Name Key Number	Provider Name Value Precision	Provider Name Value Recall	Provider Name Value F1	Provider Name Value Number	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.5109	1.0	691	0.0829	0.6804	0.6906	0.6855	598	0.9469	0.9833	0.9648	599	0.6607	0.6628	0.6617	608	0.9331	0.9362	0.9346	611	0.5872	0.6577	0.6205	558	0.9442	0.9626	0.9533	562	0.7904	0.8159	0.8030	0.9921
0.0554	2.0	1382	0.0391	0.6838	0.6906	0.6872	598	0.9950	0.9983	0.9967	599	0.6546	0.6859	0.6699	608	0.9357	0.9525	0.9440	611	0.7431	0.7724	0.7575	558	0.9231	0.9609	0.9416	562	0.8214	0.8430	0.8321	0.9926
0.0298	3.0	2073	0.0255	0.6922	0.6957	0.6939	598	0.9934	0.9983	0.9958	599	0.6505	0.6859	0.6677	608	0.9560	0.9591	0.9575	611	0.7718	0.7760	0.7739	558	0.9459	0.9644	0.9551	562	0.8332	0.8462	0.8396	0.9931
0.0213	4.0	2764	0.0208	0.6264	0.6756	0.6500	598	0.9934	0.9983	0.9958	599	0.6762	0.6974	0.6866	608	0.9645	0.9787	0.9716	611	0.7334	0.7545	0.7438	558	0.9579	0.9715	0.9647	562	0.8222	0.8459	0.8338	0.9928
0.0177	5.0	3455	0.0185	0.6672	0.6973	0.6819	598	0.9934	0.9983	0.9958	599	0.6856	0.6957	0.6906	608	0.9686	0.9591	0.9638	611	0.7778	0.7778	0.7778	558	0.9444	0.9680	0.9561	562	0.8378	0.8490	0.8434	0.9933
0.0157	6.0	4146	0.0176	0.6846	0.7007	0.6926	598	0.9934	0.9983	0.9958	599	0.6852	0.7089	0.6968	608	0.9581	0.9738	0.9659	611	0.7875	0.7903	0.7889	558	0.9786	0.9751	0.9768	562	0.8460	0.8575	0.8517	0.9936
0.0144	7.0	4837	0.0176	0.6228	0.6656	0.6435	598	0.9950	0.9983	0.9967	599	0.6667	0.6974	0.6817	608	0.9677	0.9804	0.9740	611	0.7718	0.7760	0.7739	558	0.9658	0.9555	0.9606	562	0.8275	0.8453	0.8363	0.9931
0.0134	8.0	5528	0.0172	0.6551	0.6923	0.6732	598	0.9933	0.9967	0.995	599	0.6900	0.7138	0.7017	608	0.9741	0.9853	0.9797	611	0.7715	0.7867	0.7791	558	0.9713	0.9644	0.9679	562	0.8395	0.8563	0.8478	0.9937
0.0125	9.0	6219	0.0167	0.6571	0.6890	0.6727	598	0.9934	0.9983	0.9958	599	0.7	0.7138	0.7068	608	0.9676	0.9771	0.9723	611	0.7786	0.7814	0.7800	558	0.9715	0.9698	0.9706	562	0.8425	0.8546	0.8485	0.9936
0.0118	10.0	6910	0.0170	0.6683	0.6839	0.6760	598	0.9934	0.9983	0.9958	599	0.6949	0.7155	0.7050	608	0.9804	0.9820	0.9812	611	0.7583	0.7760	0.7671	558	0.9751	0.9751	0.9751	562	0.8432	0.8549	0.8490	0.9936

Framework versions

• Transformers 4.28.0
• Pytorch 2.0.0+cu118
• Datasets 2.2.2
• Tokenizers 0.13.3