AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

TIPS: Turn-level Information-Potential Reward Shaping for Search-Augmented LLMs

Yutao Xie, Nathaniel Thomas, Nicklas Hansen, Yang Fu, Li Erran Li, Xiaolong Wang

ICLR 2026 · Reinforcement learning › RL from feedback · approach: RL for LLM reasoning · 14 citations (OpenAlex)

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 turn-level information-gain rewardPPO · Search QA avg (Qwen2.5-7B)EM: 37.28 -> 41.71Improved
add turn-level information-gain rewardPPO · Search QA avg (Qwen2.5-3B)EM: 30.15 -> 33.60Improved
add turn-level information-gain rewardPPO · Search QA avg (Qwen2.5-7B)F1: 45.07 -> 51.24Improved
add turn-level information-gain rewardPPO · 2WikiMultiHopQA (Qwen2.5-7B)EM: 32.94 -> 42.96Improved
add turn-level information-gain rewardPPO · NQ (Qwen2.5-3B)EM: 43.80 -> 43.46No clear effect
use MT-GRPO turn-level rule rewards instead of GRPOGRPO · Search QA avg (Qwen2.5-7B)EM: 28.54 -> 30.42Improved
use MT-GRPO instead of GRPOGRPO · Search QA avg (Qwen2.5-3B)EM: 30.86 -> 26.35Hurt
use a different teacher (TIPS-trained Qwen3-4B) instead of frozen policy (ablation)TIPS · Search QA avg (Qwen2.5-7B)EM: 41.7 -> 30.0Hurt
use Llama3.1-8B teacher instead of frozen policy (ablation)TIPS · Search QA avg (Qwen3-4B)EM: 48.4 -> 43.0Hurt
fix the teacher (refresh N=500) rather than refresh every 200 steps (ablation)TIPS · Search QA avgEMHurt
use large shaping target band (0.3-1.0) instead of medium (ablation)TIPS · Search QA avg (Qwen3-4B)EM gain over PPO: +7.3% -> -3.4%Hurt
use small shaping target band (0.001-0.05) (ablation)TIPS · Search QA avg (Qwen2.5-7B)EM gain over PPO: +11.9% -> 0% (crashed)Hurt

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