AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

A Rolling Stone Gathers No Moss: Adaptive Policy Optimization for Stable Self-Evaluation in Large Multimodal Models

Wenkai Wang, Hongcan Guo, Zheqi Lv, Shengyu Zhang

AAAI 2026 · Reinforcement learning › RL from feedback · approach: RL for LLM reasoning · 0 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 Adaptive Reward Model (ablation)AdaPO (Qwen2.5-VL-7B) · 8 benchmarks avg (Qwen2.5-VL-7B)Avg acc@t2: 53.8 -> 60.2Improved
add Adaptive Reward Model (ablation)AdaPO (Qwen2.5-VL-7B) · 8 benchmarks avg (Qwen2.5-VL-7B)Avg acc@t1: 55.1 -> 57.6Improved
add Reward Aware KL (ablation)AdaPO (Qwen2.5-VL-7B) · 8 benchmarks avg (Qwen2.5-VL-7B)Avg acc@t1: 56.2 -> 57.6Improved
add Reward Aware KL (ablation)AdaPO (Qwen2.5-VL-7B) · 8 benchmarks avg (Qwen2.5-VL-7B)Avg acc@t2: 58.6 -> 60.2Improved
add hybrid data filtering (ablation)AdaPO (Qwen2.5-VL-7B) · 8 benchmarks avg (Qwen2.5-VL-7B)Avg acc@t2: 57.5 -> 60.2Improved
add hybrid data filtering (ablation)AdaPO (Qwen2.5-VL-7B) · 8 benchmarks avg (Qwen2.5-VL-7B)Avg acc@t1: 54.3 -> 57.6Improved
prompt base model for self-evaluation (no training)Qwen2.5-VL-7B · MathVistaacc@t2 vs acc@t1: 68.2 -> 60.0Hurt
fixed-reward GRPO for self-evaluationQwen2.5-VL-7B · AI2Dacc@t2 vs acc@t1: 80.5 -> 80.9No clear effect
replace GRPO with AdaPOQwen2.5-VL-7B · AI2Dacc@t2: 80.9 -> 85.5Improved
replace SCoRe staged RL with AdaPOQwen2.5-VL-7B · MMMUacc@t1: 39.4 -> 54.4Improved
replace GRPO with AdaPOQwen2.5-VL-7B · WeMathacc@t2: 65.7 -> 73.4Improved
apply SFT or STaR for self-evaluationQwen2.5-VL-7B · WeMathacc@t2 vs acc@t1: 61.9 -> 51.1Hurt

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