ResearchBenchmarksMulti-Hop Question Answering on LV-Eval (EM/F1)For AI assistantsMarkdownContent indexMulti-Hop Question Answering on LV-Eval (EM/F1)Follow10.5EMHELP6.347.428.59.58Feb 24, 2026Evaluation resultsMethodMethodLinksEMF1HELPBackbone=Qwen3-30B-A3B...Backbone=Qwen3-30B-A3B-Instruct-25072026.0210.512.9HippoRAG2Backbone=Qwen3-30B-A3B...Backbone=Qwen3-30B-A3B-Instruct-25072026.027.269.67LinearRAGBackbone=Qwen3-30B-A3B...Backbone=Qwen3-30B-A3B-Instruct-25072026.026.510.7