Personal Information Extraction on Physician benchmark leaderboard · SOTA2 Research

Personal Information Extraction on Physician

100Email Address Accuracygpt-4

Evaluation Results

Method	Links
gpt-42024.08		100	1	100	1	100	1	95	0.93	57	0.38	63	0.49	100	1	100	1
gpt-3.5-turbo2024.08		100	1	100	1	100	1	64	0.63	33	0.21	36	0.25	100	1	100	1
vicuna-13b-v1.32024.08		99	0.99	100	1	99	0.67	33	0.28	35	0.38	18	0.22	99	0.67	99	0.67
gemini-pro2024.08		97	0.97	100	1	98	0.98	92	0.91	43	0.26	57	0.41	98	0.98	97	0.97
text-bison-0012024.08		85	0.85	87	0.87	79	0.77	89	0.99	31	0.21	49	0.41	96	0.96	97	0.97
vicuna-7b-v1.32024.08		55	0.55	95	0.95	68	0.33	26	0.58	14	0.14	11	0.08	62	0.03	59	0.05
chat-bison-0012024.08		54	0.54	90	0.9	18	0.18	94	0.94	32	0.12	25	0.07	35	0.35	28	0.28
llama-2-7b-chat-hf2024.08		50	0.5	25	0.25	67	0.67	40	0.4	34	0.04	19	0.07	25	0.25	25	0.25
flan-ul22024.08		48	0.48	83	0.83	48	0.48	99	0.99	11	0.02	16	0.06	28	0.28	25	0.25
internlm-chat-7b2024.08		34	0.34	47	0.47	43	0.43	81	0.81	26	0.1	20	0.05	21	0.21	14	0.14