# Multihop Question Answering

Retrieval-augmented multihop question answering suffers from noisy context, and CLaRa's compression of retrieved documents into dense latent memory makes the structural relations between documents harder to exploit. GraphCLaRa places a roughly 1.3M-parameter graph reranker before the frozen CLaRa reader: it builds a query-aware document graph, propagates evidence with a graph attention network, and adaptively decides how many documents to pass on. Across four QA benchmarks it raises answer F1 on three, by 6.6 on 2Wiki, 6.4 on MuSiQue and 2.5 on NQ, while using fewer documents, and trails by 1.1 F1 on the retrieval-saturated HotpotQA.

Retrieval-augmented multihop question answering suffers from noisy context, and CLaRa's compression of retrieved documents into dense latent memory makes the structural relations between documents harder to exploit. GraphCLaRa places a roughly 1.3M-parameter graph reranker before the frozen CLaRa reader: it builds a query-aware document graph, propagates evidence with a graph attention network, and adaptively decides how many documents to pass on. Across four QA benchmarks it raises answer F1 on three, by 6.6 on 2Wiki, 6.4 on MuSiQue and 2.5 on NQ, while using fewer documents, and trails by 1.1 F1 on the retrieval-saturated HotpotQA.

Retrieval-augmented generation does not improve every task, since irrelevant evidence can distract a generator, and measuring each candidate retriever's gain directly repeats generation for every configuration. EviCast splits a retriever's net utility into the generator's response to controlled evidence states and that retriever's offline distribution over those states; the response is measured once on a small calibration set and reused. On 14 HotpotQA retrieval configurations, one calibration used 96 generation calls, made 14 of 14 correct deployment decisions with a 2.75 F1-point forecast error, and cut diagnostic generation calls by 80% relative to per-configuration direct measurement.

Retrieval-augmented generation does not improve every task, since irrelevant evidence can distract a generator, and measuring each candidate retriever's gain directly repeats generation for every configuration. EviCast splits a retriever's net utility into the generator's response to controlled evidence states and that retriever's offline distribution over those states; the response is measured once on a small calibration set and reused. On 14 HotpotQA retrieval configurations, one calibration used 96 generation calls, made 14 of 14 correct deployment decisions with a 2.75 F1-point forecast error, and cut diagnostic generation calls by 80% relative to per-configuration direct measurement.