AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

GPS: Directed Acyclic Graph guided Proactive Information Seeking in Large Language Models

Ruiqing Li, Yifeng Xu, Xinke Jiang, Zhibang Yang, Xinyu Ma, Yue Fang, Junfeng Zhao, Yasha Wang, Xu Chu

ICLR 2026 · Large language models › Retrieval and knowledge · approach: Knowledge-augmented question answering

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
add clarification-oriented reinforcement learning (ablation)GPS (Qwen2.5-7B-Instruct) · SyntheticSuccess Rate (%): 52.2 -> 60.2Improved
add clarification-oriented reinforcement learning (ablation)GPS (Qwen2.5-7B-Instruct) · SyntheticWeighted Clarification Turns: 5.57 -> 4.59Improved
add clarification-oriented reinforcement learning (ablation)GPS (Qwen2.5-7B-Instruct) · ConditionalQASuccess Rate (%): 67.7 -> 73.4Improved
add clarification-oriented reinforcement learning (ablation)GPS (Qwen2.5-7B-Instruct) · ConditionalQAWeighted Clarification Turns: 3.63 -> 2.91Improved
add efficiency reward (ablation)GPS (Qwen2.5-7B-Instruct) · SyntheticSuccess Rate (%): 59.0 -> 60.2Improved
add efficiency reward (ablation)GPS (Qwen2.5-7B-Instruct) · SyntheticWeighted Clarification Turns: 5.06 -> 4.59Improved
add efficiency reward (ablation)GPS (Qwen2.5-7B-Instruct) · ConditionalQASuccess Rate (%): 70.7 -> 73.4Improved
add structural quality reward (ablation)GPS (Qwen2.5-7B-Instruct) · SyntheticSuccess Rate (%): 56.1 -> 60.2Improved
add structural quality reward (ablation)GPS (Qwen2.5-7B-Instruct) · SyntheticWeighted Clarification Turns: 5.32 -> 4.59Improved
add structural quality reward (ablation)GPS (Qwen2.5-7B-Instruct) · ConditionalQASuccess Rate (%): 70.3 -> 73.4Improved
add dynamic traversal (ablation)GPS (Qwen2.5-7B-Instruct) · SyntheticWeighted Clarification Turns: 5.19 -> 4.59Improved
add dynamic traversal (ablation)GPS (Qwen2.5-7B-Instruct) · ConditionalQAWeighted Clarification Turns: 3.63 -> 2.91Improved

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