Multi-task Graph-based Question Answering and Generation on GraphBench Medical 1.0 (test) benchmark leaderboard · SOTA2 Research

Multi-task Graph-based Question Answering and Generation on GraphBench Medical 1.0 (test)

73.95Fact Retrieval ACCCogitoRAG

Evaluation Results

Method	Links
CogitoRAG2026.02		73.95	39.34	74.43	37.11	74.77	45.29	61.99	32.68	30.91	71.29
HippoRAG22026.02		66.28	36.69	61.98	36.97	63.08	46.13	68.05	58.78	51.54	64.85
RAG2026.02		64.73	30.75	58.64	15.57	65.75	78.54	60.61	36.74	58.72	62.43
RAG2026.02		63.72	29.21	57.61	13.98	63.72	77.34	58.94	35.88	57.87	61
LightRAG2026.02		63.32	37.19	61.32	24.98	63.14	51.16	67.91	78.76	51.58	63.92
Fast-GraphRAG2026.02		60.93	31.04	61.73	21.37	67.88	52.07	65.93	56.07	44.73	64.12
Lazy-GraphRAG2026.02		60.25	31.66	47.82	22.68	57.28	55.92	62.22	30.95	43.79	56.89
HippoRAG2026.02		56.14	20.95	55.87	13.57	59.86	62.73	64.43	69.21	65.56	59.08
RAPTOR2026.02		54.07	17.93	53.2	11.73	58.73	78.28	62.38	59.98	63.63	57.1
MS-GraphRAG2026.02		38.63	26.8	47.04	21.99	41.87	22.98	53.11	32.65	39.42	45.16