Restricting RL training retrieval to ground-truth-related documents improves a search agent's multi-hop QA exact match from 0.206 to 0.297 over Search-R1, when tested on the full corpus.
Question answering tasks can be categorized into two sub-classes: general question answering
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Search-GRT: Guided Retrieval Training of Search Agents to Optimize for Complex Question Answering
Restricting RL training retrieval to ground-truth-related documents improves a search agent's multi-hop QA exact match from 0.206 to 0.297 over Search-R1, when tested on the full corpus.