AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

RLMR: Reinforcement Learning with Mixed Rewards for Creative Writing

JianXing Liao, Tian Zhang, Xiao Feng, Yusong Zhang, Haorui Wang, Bosi Wen, Ziying Wang, Runzhi Shi

AAAI 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
train with writing reward model onlyQwen2.5-32B GRPO · ComplexBenchaccuracy: 74.78% -> 68.42%Hurt
train with writing reward model onlyQwen2.5-32B GRPO · WritingBenchscore: 6.14 -> 7.05Improved
train with writing reward model onlyQwen2.5-32B GRPO · IFEvalaccuracy: 83.36% -> 80.41%Hurt
train with constraint verification onlyQwen2.5-32B GRPO · WriteEvalwin rate: 3.93% -> 1.24%Hurt
train with constraint verification onlyQwen2.5-32B GRPO · ComplexBenchaccuracy: 74.78% -> 83.94%Improved
use RLMR dynamic penalty instead of fixed linear weightingQwen2.5-32B · WritingBenchscore: 7.13 -> 7.93Improved
use RLMR instead of linear weightingQwen2.5-32B · ComplexBenchaccuracy: 73.91% -> 79.04%Improved
use RLMR instead of linear weightingQwen2.5-72B · WriteEvalwin rate: 10.22% -> 17.18%Improved
use RLMR instead of linear weightingQwen3-8B · WriteEvalwin rate: 26.64% -> 31.69%Improved
use RLMR instead of linear weightingDeepSeek-R1-Distill-Llama-8B · ComplexBenchaccuracy: 53.91% -> 52.35%Hurt
use RLMR instead of linear weightingDeepSeek-R1-Distill-Llama-8B · WritingBenchscore: 5.68 -> 7.41Improved
add dynamic sampling (DAPO-style filtering) to RLMR (ablation)Qwen2.5-32B · WritingBenchscore: 7.34 -> 7.93Improved

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