AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

SERL: Self-Examining Reinforcement Learning on Open-Domain

Weixuan Ou, Yanzhao Zheng, Shuoshuo Sun, Wei Zhang, Baohua Dong, Hangcheng Zhu, Ruohui Huang, Gang Yu, Pengwei Yan, Yifan Qiao

AAAI 2026 · Reinforcement learning › RL from feedback · approach: RL for LLM reasoning · 2 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 reward for Judge (self-consistency) (ablation)SERL (Qwen3-8B) · Summarizationwin rate: -9.34% when removedImproved
add reward for Judge (self-consistency) (ablation)SERL (Qwen3-8B) · Open Writingwin rate: -10.66% when removedImproved
add reward for Judge (self-consistency) (ablation)SERL (Qwen3-8B) · AlpacaEval 2.0LC win rate: 59.90% -> 54.50% when removedImproved
add reward for Actor (Copeland win-rate ranking) (ablation)SERL (Qwen3-8B) · Summarizationwin rate: -35.34% when removedImproved
add reward for Actor (Copeland win-rate ranking) (ablation)SERL (Qwen3-8B) · Open Writingwin rate: -20.34% when removedImproved
add reward for Actor (Copeland win-rate ranking) (ablation)SERL (Qwen3-8B) · AlpacaEval 2.0LC win rate: 59.90% -> 51.47% when removedImproved
add position bias mitigation (PBMM) (ablation)SERL (Qwen3-8B) · Summarizationwin rate: -15.00% when removedImproved
add position bias mitigation (PBMM) (ablation)SERL (Qwen3-8B) · Open Writingwin rate: -2.66% when removedImproved
add position bias mitigation (PBMM) (ablation)SERL (Qwen3-8B) · AlpacaEval 2.0LC win rate: 59.90% -> 54.01% when removedImproved
add length control module (LCM) (ablation)SERL (Qwen3-8B) · Summarization trainingaverage output length: surpassing 1,000 tokens without LCM vs around 600 tokens with LCMImproved
train Qwen3-8B with SERLQwen3-8B · AlpacaEval 2.0LC win rate: 52.37% -> 59.90%Improved
train Qwen3-8B with SERLQwen3-8B · AlpacaEval 2.0win rate: 55.07% -> 69.88%Improved

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