MuSiQue
A standard benchmark dataset in the multi-hop question answering domain, focused on complex question answering tasks requiring reasoning across multiple documents. It is distinguished by its question complexity design, intended to probe deep multi-step reasoning capabilities in language models.
Timeline (last 90 days)
Multi-hop QA is a class of questions that require reasoning across multiple independent evidence fragments to answer, with mainstream benchmarks including HotpotQA, MuSiQue, and 2WikiMultiHopQA.
The experiment was conducted on 400 samples drawn from three multi-hop QA datasets: MuSiQue, HotpotQA, and 2WikiMultiHopQA.
All Facts (2)
Multi-hop QA is a class of questions that require reasoning across multiple independent evidence fragments to answer, with mainstream benchmarks including HotpotQA, MuSiQue, and 2WikiMultiHopQA.
50%UnverifiedThe experiment was conducted on 400 samples drawn from three multi-hop QA datasets: MuSiQue, HotpotQA, and 2WikiMultiHopQA.
50%