AI Paper Atlas

Research map / Large language models / Reasoning and chain of thought

CP-Search: A Chain Progressive Search Training Framework Incentivizing the Cognitive Behaviors for Searching in LLMs

Zehua Wang, Shipeng Li, Buzhou Tang

AAAI 2026 · Large language models › Reasoning and chain of thought · approach: RL with verifiable rewards for reasoning · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes CP-Search, a chain-progressive search training framework that uses reinforcement learning to incentivize cognitive search behaviors in LLMs for multi-hop retrieval-augmented reasoning.

Problem

Existing RAG methods model reasoning paths poorly for complex multi-hop reasoning, and RL-based reasoning training has been applied to search but the abstract is truncated before details of its limitations are stated.

Approach

A chain-progressive search training framework that incentivizes cognitive search behaviors in LLMs during RL training, aimed at improving multi-hop reasoning with retrieval; specific mechanism not stated in the available abstract.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add cognitive-behavior SFT stage (ablation)CP-Search · 2WikiMultiHopQAACCR: 56.0 -> 72.4Improved
add cognitive-behavior SFT stage (ablation)CP-Search · MuSiQueACCR: 21.6 -> 32.2Improved
add procedural-supervision RL stage after SFT (ablation)CP-Search · 2WikiMultiHopQAACCR: 64.2 -> 72.4Improved
add procedural-supervision RL stage after SFT (ablation)CP-Search · MuSiQueF1: 26.5 -> 36.8Improved
replace GRPO with CPGRPO procedural reward (ablation)CP-Search · 2WikiMultiHopQAACCR: 68.8 -> 72.4Improved
replace GRPO with CPGRPO procedural reward (ablation)CP-Search · MuSiQueF1: 35.2 -> 36.8Improved
CP-Search vs R1-Searcher on HotpotQAQwen2.5-7B-Instruct · HotpotQAAVG: 70.2 -> 69.2Hurt
CP-Search vs R1-Searcher on 2WikiQwen2.5-7B-Instruct · 2WikiMultiHopQAAVG: 64.3 -> 72.2Improved
RL stage after SFT: effect on capability (ablation)CP-Search · 2Wiki, MuSiQuepass@kMixed

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