AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Multimodal LLM-assisted Evolutionary Search for Programmatic Control Policies

Qinglong Hu, Tong Xialiang, Mingxuan Yuan, Fei Liu, Zhichao Lu, Qingfu Zhang

ICLR 2026 · Reinforcement learning › RL from feedback · approach: RL fine-tuning of generative models · 9 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 visual behavioral-evidence analysis to EoH (MLES vs EoH)MLES-EoH · Car Racing testingscore: 79.286 -> 96.358Improved
add visual behavioral-evidence analysis to EoH (MLES vs EoH)MLES-EoH · Lunar Lander testingscore: 0.776 -> 0.819Mixed
compare MLES with PPO on test instancesMLES · Lunar Lander testingscore: 0.846 -> 0.819Hurt
compare MLES with PPO on Car Racing testMLES · Car Racing testingscore: 94.546 -> 96.358Improved
remove multimodal modification operator M1_M (ablation)MLES · Lunar Landerscore: 0.997 -> 1.090Improved
remove multimodal modification operator M2_M (ablation)MLES · Car Racingscore: 84.77 -> 98.70Improved
add exploration operator E1 (ablation)MLES · Lunar Landerscore: 1.093 -> 1.090No clear effect
add exploration operator E1 (ablation)MLES · Car Racingscore: 87.53 -> 98.70Improved
provide an initial seed policy (GPT-4o-mini) (ablation)MLES · Lunar Landerscore @ 2k: 0.667 -> 1.090Improved
use GPT-5-mini instead of GPT-4o-mini without seeds (ablation)MLES · Lunar Lander cold startscore @ 2k: 0.667 -> 1.013Improved
use GPT-5-mini instead of GPT-4o-mini with seeds (ablation)MLES · Lunar Landerscore @ 2k: 1.090 -> 1.065Hurt
add image evidence to M1 operator (ablation)M1 · Lunar Lander operator testbest policy improvement: +1.18% -> +4.39%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