AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Reinforcement Unlearning via Group Relative Policy Optimization

Efstratios Zaradoukas, Bardh Prenkaj, Gjergji Kasneci

ICLR 2026 · Reinforcement learning › RL from feedback · approach: RL for LLM reasoning · 3 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
use GRPO with intrinsic verifiable reward (PURGE) instead of base modelPhi-3-Mini · RWKU Forget Set AArecall: -12.02%Improved
PURGE vs NPO on forget qualityPURGE · RWKU Forget Set FBrecall: .428 vs .296Hurt
PURGE forget set construction vs DPOPURGE · Forget set tokenstokens per target: x46 fewer than DPO (615,585)Improved
PURGE forget set construction vs GA/NPOPURGE · Forget set tokenstokens per target: x22 fewer than GA and NPO (298,067)Improved
PURGE vs base on fluencyPURGE · RWKU FLUfluency: 133.58 -> 140.90Improved
PURGE vs base on utility GAPURGE · RWKU Utility GAaccuracy: -3.62%Hurt
increase Qwen-2.5 model size 0.5B to 3B (ablation)PURGE · RWKU forget probesrecallHurt
scale Qwen-2.5 size on utility (ablation)PURGE · RWKU utilityutility metricsNo clear effect
swap NER-constructing LLM (Claude, Deepseek etc.) (ablation)PURGE · RWKU Stephen Kingforget/utility metricsNo clear effect

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