AI Paper Atlas

Research map / Large language models / Agents, code and math

Enhancing Agentic Search via Data Synthesis on Hierarchical Constraint Satisfaction

Ziyi Xia, Kun Luo, Hongjin Qian, Siqi Bao, Zheng Liu

ICLR 2026 · Large language models › Agents, code and math · approach: LLMs for optimization and heuristics

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train on InfoSeek data (SFT then GRPO) instead of search-agent baselines trained on NQ/HotpotQAQwen2.5-3B-Instruct · 7 QA benchmarks (average)exact match: 40.6 -> 42.7 (vs InForage-3B)Improved
InfoSeeker-3B vs baselines on single-hop NQ and TQAInfoSeeker-3B · NQ, TriviaQAexact match: NQ 41.7 vs AutoRefine 43.6; TQA 56.1 vs ZeroSearch 61.5Mixed
InfoSeeker-3B vs strongest baselines on multi-hop setsInfoSeeker-3B · HotpotQA, 2Wikiexact match: HQA 40.9 -> 44.6; 2Wiki 42.8 -> 50.0Improved
train 3B model on InfoSeekInfoSeeker-3B · BrowseComp-Plusaccuracy: 3.9 -> 15.3 (vs SearchR1-32B)Improved
InfoSeeker-3B vs GPT-5 on BrowseComp-PlusInfoSeeker-3B · BrowseComp-Plusaccuracy: 55.9 vs 15.3Hurt
add InfoSeek SFT stage before RL (ablation)Qwen2.5-3B-Instruct · 7 QA benchmarksexact matchImproved
scale model from 3B to 7B (ablation)InfoSeeker · 7 QA benchmarksexact matchImproved
train on InfoSeek instead of only NQ+HotpotQA (ablation)InfoSeeker-3B · BrowseComp-Plusaccuracy and search callsImproved
restrict InfoSeek to fewer than five vertices (ablation)InfoSeeker-3B · BrowseComp-Plusaccuracy and search callsImproved
increase training data scale (1K, 5K, 18K) (ablation)InfoSeeker-3B · BrowseComp-Plusaccuracy and search callsImproved
increase vertices per question (ablation)InfoSeek dataset · InfoSeek difficultyfailure rate of Qwen2.5-72B: 88.1 -> 94.1Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy