lmv2-g-bnkstm-994-doc-09-10

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: 0.0926
• Account Number Precision: 0.8889
• Account Number Recall: 0.9014
• Account Number F1: 0.8951
• Account Number Number: 142
• Bank Name Precision: 0.7993
• Bank Name Recall: 0.8484
• Bank Name F1: 0.8231
• Bank Name Number: 277
• Cust Address Precision: 0.8563
• Cust Address Recall: 0.8827
• Cust Address F1: 0.8693
• Cust Address Number: 162
• Cust Name Precision: 0.9181
• Cust Name Recall: 0.9290
• Cust Name F1: 0.9235
• Cust Name Number: 169
• Ending Balance Precision: 0.7706
• Ending Balance Recall: 0.7892
• Ending Balance F1: 0.7798
• Ending Balance Number: 166
• Starting Balance Precision: 0.9051
• Starting Balance Recall: 0.8720
• Starting Balance F1: 0.8882
• Starting Balance Number: 164
• Statement Date Precision: 0.8817
• Statement Date Recall: 0.8765
• Statement Date F1: 0.8791
• Statement Date Number: 170
• Overall Precision: 0.8531
• Overall Recall: 0.8688
• Overall F1: 0.8609
• Overall Accuracy: 0.9850

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: 4e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Account Number Precision	Account Number Recall	Account Number F1	Account Number Number	Bank Name Precision	Bank Name Recall	Bank Name F1	Bank Name Number	Cust Address Precision	Cust Address Recall	Cust Address F1	Cust Address Number	Cust Name Precision	Cust Name Recall	Cust Name F1	Cust Name Number	Ending Balance Precision	Ending Balance Recall	Ending Balance F1	Ending Balance Number	Starting Balance Precision	Starting Balance Recall	Starting Balance F1	Starting Balance Number	Statement Date Precision	Statement Date Recall	Statement Date F1	Statement Date Number	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.7648	1.0	795	0.2550	0.8514	0.4437	0.5833	142	0.6229	0.5307	0.5731	277	0.5650	0.7778	0.6545	162	0.6682	0.8698	0.7558	169	0.0	0.0	0.0	166	0.0	0.0	0.0	164	0.6040	0.3588	0.4502	170	0.6370	0.4352	0.5171	0.9623
0.1725	2.0	1590	0.1128	0.6067	0.7606	0.675	142	0.7294	0.7978	0.7621	277	0.8150	0.8704	0.8418	162	0.8966	0.9231	0.9096	169	0.7786	0.6566	0.7124	166	0.7576	0.7622	0.7599	164	0.8509	0.8059	0.8278	170	0.7705	0.7976	0.7838	0.9816
0.0877	3.0	2385	0.0877	0.7857	0.9296	0.8516	142	0.7872	0.8014	0.7943	277	0.7709	0.8519	0.8094	162	0.8827	0.9349	0.9080	169	0.7673	0.7349	0.7508	166	0.8313	0.8415	0.8364	164	0.7716	0.8941	0.8283	170	0.7985	0.8496	0.8233	0.9830
0.0564	4.0	3180	0.0826	0.8503	0.8803	0.8651	142	0.7566	0.8303	0.7917	277	0.7895	0.8333	0.8108	162	0.8824	0.8876	0.8850	169	0.7049	0.7771	0.7393	166	0.7717	0.8659	0.8161	164	0.8363	0.8412	0.8387	170	0.7925	0.8432	0.8171	0.9828
0.0402	5.0	3975	0.0889	0.8815	0.8380	0.8592	142	0.7758	0.7870	0.7814	277	0.8266	0.8827	0.8537	162	0.8983	0.9408	0.9191	169	0.6378	0.7108	0.6724	166	0.8707	0.7805	0.8232	164	0.8508	0.9059	0.8775	170	0.8124	0.8312	0.8217	0.9837
0.0332	6.0	4770	0.0864	0.7778	0.9366	0.8498	142	0.8175	0.8412	0.8292	277	0.8704	0.8704	0.8704	162	0.9167	0.9112	0.9139	169	0.7702	0.7470	0.7584	166	0.8424	0.8476	0.8450	164	0.8728	0.8882	0.8805	170	0.8366	0.86	0.8481	0.9846
0.0285	7.0	5565	0.0858	0.7516	0.8310	0.7893	142	0.8156	0.8303	0.8229	277	0.8373	0.8580	0.8476	162	0.9133	0.9349	0.9240	169	0.8288	0.7289	0.7756	166	0.8144	0.8293	0.8218	164	0.8353	0.8353	0.8353	170	0.8279	0.8352	0.8315	0.9840
0.027	8.0	6360	0.1033	0.8841	0.8592	0.8714	142	0.7695	0.8556	0.8103	277	0.7816	0.8395	0.8095	162	0.9075	0.9290	0.9181	169	0.8538	0.6687	0.75	166	0.8861	0.8537	0.8696	164	0.8492	0.8941	0.8711	170	0.8373	0.844	0.8406	0.9837
0.0237	9.0	7155	0.0922	0.8792	0.9225	0.9003	142	0.8262	0.8412	0.8336	277	0.8421	0.8889	0.8649	162	0.8983	0.9408	0.9191	169	0.8113	0.7771	0.7938	166	0.7641	0.9085	0.8301	164	0.8466	0.8765	0.8613	170	0.8358	0.8752	0.8550	0.9850
0.023	10.0	7950	0.0935	0.8493	0.8732	0.8611	142	0.7848	0.8556	0.8187	277	0.8246	0.8704	0.8468	162	0.9080	0.9349	0.9213	169	0.8133	0.7349	0.7722	166	0.8867	0.8110	0.8471	164	0.8735	0.8529	0.8631	170	0.8419	0.848	0.8450	0.9841
0.0197	11.0	8745	0.0926	0.8889	0.9014	0.8951	142	0.7993	0.8484	0.8231	277	0.8563	0.8827	0.8693	162	0.9181	0.9290	0.9235	169	0.7706	0.7892	0.7798	166	0.9051	0.8720	0.8882	164	0.8817	0.8765	0.8791	170	0.8531	0.8688	0.8609	0.9850
0.0193	12.0	9540	0.1035	0.7514	0.9366	0.8339	142	0.8127	0.8773	0.8438	277	0.8103	0.8704	0.8393	162	0.9405	0.9349	0.9377	169	0.6983	0.7530	0.7246	166	0.8011	0.8841	0.8406	164	0.8462	0.9059	0.8750	170	0.8081	0.8792	0.8421	0.9836
0.0166	13.0	10335	0.1077	0.8889	0.8451	0.8664	142	0.8062	0.8412	0.8233	277	0.7953	0.8395	0.8168	162	0.8786	0.8994	0.8889	169	0.8069	0.7048	0.7524	166	0.8167	0.8963	0.8547	164	0.8671	0.8824	0.8746	170	0.8333	0.844	0.8386	0.9836
0.016	14.0	11130	0.1247	0.8521	0.8521	0.8521	142	0.8456	0.8303	0.8379	277	0.8050	0.7901	0.7975	162	0.9167	0.9112	0.9139	169	0.8392	0.7229	0.7767	166	0.8521	0.8780	0.8649	164	0.9262	0.8118	0.8652	170	0.8611	0.828	0.8442	0.9836
0.0153	15.0	11925	0.1030	0.8280	0.9155	0.8696	142	0.7637	0.8051	0.7838	277	0.8452	0.8765	0.8606	162	0.9337	0.9172	0.9254	169	0.7551	0.6687	0.7093	166	0.8616	0.8354	0.8483	164	0.8287	0.8824	0.8547	170	0.8252	0.8384	0.8317	0.9834
0.0139	16.0	12720	0.0920	0.8075	0.9155	0.8581	142	0.7735	0.8628	0.8157	277	0.7663	0.8704	0.8150	162	0.8870	0.9290	0.9075	169	0.7647	0.7831	0.7738	166	0.8571	0.8780	0.8675	164	0.6630	0.7176	0.6893	170	0.7857	0.8504	0.8167	0.9832
0.0124	17.0	13515	0.1057	0.8013	0.8521	0.8259	142	0.8087	0.8087	0.8087	277	0.7663	0.8704	0.8150	162	0.9186	0.9349	0.9267	169	0.8322	0.7169	0.7702	166	0.8563	0.8720	0.8640	164	0.8603	0.9059	0.8825	170	0.8327	0.848	0.8403	0.9829
0.0135	18.0	14310	0.1001	0.8323	0.9085	0.8687	142	0.8363	0.8484	0.8423	277	0.8494	0.8704	0.8598	162	0.8462	0.9112	0.8775	169	0.7925	0.7590	0.7754	166	0.8286	0.8841	0.8555	164	0.8686	0.8941	0.8812	170	0.8368	0.8656	0.8510	0.9839
0.0125	19.0	15105	0.1200	0.8562	0.8803	0.8681	142	0.8	0.8520	0.8252	277	0.7705	0.8704	0.8174	162	0.8864	0.9231	0.9043	169	0.7716	0.7530	0.7622	166	0.8642	0.8537	0.8589	164	0.85	0.9	0.8743	170	0.8252	0.8608	0.8426	0.9843
0.0098	20.0	15900	0.1097	0.8993	0.8803	0.8897	142	0.7933	0.8592	0.8250	277	0.8144	0.8395	0.8267	162	0.8641	0.9408	0.9008	169	0.82	0.7410	0.7785	166	0.8704	0.8598	0.8650	164	0.8876	0.8824	0.8850	170	0.8434	0.8576	0.8505	0.9846
0.0128	21.0	16695	0.1090	0.8993	0.8803	0.8897	142	0.8294	0.8773	0.8526	277	0.8107	0.8457	0.8278	162	0.8678	0.8935	0.8805	169	0.8133	0.7349	0.7722	166	0.8218	0.8720	0.8462	164	0.8889	0.8471	0.8675	170	0.8446	0.852	0.8483	0.9838
0.01	22.0	17490	0.1280	0.9	0.8239	0.8603	142	0.7848	0.8556	0.8187	277	0.8057	0.8704	0.8368	162	0.8674	0.9290	0.8971	169	0.7595	0.7229	0.7407	166	0.8412	0.8720	0.8563	164	0.7989	0.8882	0.8412	170	0.8169	0.8528	0.8344	0.9832
0.0096	23.0	18285	0.1023	0.8889	0.9014	0.8951	142	0.8041	0.8448	0.8239	277	0.8253	0.8457	0.8354	162	0.8415	0.9112	0.875	169	0.7683	0.7590	0.7636	166	0.8118	0.8415	0.8263	164	0.7979	0.8824	0.8380	170	0.8170	0.8536	0.8349	0.9843
0.0088	24.0	19080	0.1172	0.8649	0.9014	0.8828	142	0.8298	0.8448	0.8372	277	0.7816	0.8395	0.8095	162	0.8674	0.9290	0.8971	169	0.7257	0.7651	0.7449	166	0.8136	0.8780	0.8446	164	0.8229	0.8471	0.8348	170	0.8155	0.856	0.8353	0.9829
0.0083	25.0	19875	0.1090	0.7401	0.9225	0.8213	142	0.8363	0.8484	0.8423	277	0.8057	0.8704	0.8368	162	0.8889	0.8994	0.8941	169	0.8176	0.7289	0.7707	166	0.7609	0.8537	0.8046	164	0.8488	0.8588	0.8538	170	0.8150	0.8528	0.8335	0.9830
0.0105	26.0	20670	0.1191	0.7241	0.8873	0.7975	142	0.7468	0.8412	0.7912	277	0.8161	0.8765	0.8452	162	0.8254	0.9231	0.8715	169	0.7384	0.7651	0.7515	166	0.8333	0.8537	0.8434	164	0.8378	0.9118	0.8732	170	0.7853	0.8632	0.8224	0.9814
0.0103	27.0	21465	0.1125	0.8378	0.8732	0.8552	142	0.8566	0.8628	0.8597	277	0.8046	0.8642	0.8333	162	0.8764	0.9231	0.8991	169	0.8289	0.7590	0.7925	166	0.8466	0.8415	0.8440	164	0.8929	0.8824	0.8876	170	0.8502	0.8584	0.8543	0.9847
0.0081	28.0	22260	0.1301	0.8601	0.8662	0.8632	142	0.8489	0.8520	0.8505	277	0.8225	0.8580	0.8399	162	0.8870	0.9290	0.9075	169	0.8067	0.7289	0.7658	166	0.8625	0.8415	0.8519	164	0.8613	0.8765	0.8688	170	0.8504	0.8504	0.8504	0.9850
0.0079	29.0	23055	0.1458	0.9104	0.8592	0.8841	142	0.8185	0.8303	0.8244	277	0.7730	0.7778	0.7754	162	0.8191	0.9112	0.8627	169	0.8013	0.7530	0.7764	166	0.8304	0.8659	0.8478	164	0.8941	0.8941	0.8941	170	0.8321	0.8408	0.8365	0.9834
0.0084	30.0	23850	0.1264	0.8435	0.8732	0.8581	142	0.8328	0.8628	0.8475	277	0.8256	0.8765	0.8503	162	0.9023	0.9290	0.9155	169	0.8531	0.7349	0.7896	166	0.8598	0.8598	0.8598	164	0.8757	0.8706	0.8732	170	0.8543	0.8584	0.8563	0.9848

Framework versions

• Transformers 4.22.0.dev0
• Pytorch 1.12.1+cu113
• Datasets 2.2.2
• Tokenizers 0.12.1