AI Paper Atlas

Research map / Large language models / Agents, code and math

Grounding LLMs in Scientific Discovery via Embodied Actions

Bo Zhang, Jinfeng Zhou, Yuxuan Chen, Jianing Yin, Minlie Huang, Hongning Wang

ICML 2026 · Large language models › Agents, code and math · approach: Agent instruction-following benchmarks · 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 Reflective Decision Maker (ablation)EmbodiedAct (GPT-5.2) · EngDesign Coreaverage score: 65.0 -> 70.6Improved
add Runtime Perception engine (ablation)EmbodiedAct (GPT-5.2) · EngDesign Coreaverage score: 56.5 -> 70.6Improved
add Runtime Perception engine (ablation)EmbodiedAct (GPT-5-mini) · EngDesign Coreaverage score: 48.5 -> 59.3Improved
add both Reflective Decision Maker and Runtime Perception (ablation)EmbodiedAct (GPT-5.2) · EngDesign Coreaverage score: 51.5 -> 70.6Improved
add Reflective Decision Maker (ablation)EmbodiedAct (GPT-5-mini) · SciBench-107accuracy (%): 48.0 -> 49.5Improved
add Runtime Perception engine (ablation)EmbodiedAct (GPT-5.2) · SciBench-107accuracy (%): 50.0 -> 54.2Improved
use EmbodiedAct instead of CodeActGPT-5.2 · EngDesign Coreaverage score: 55.4 -> 70.6Improved
use EmbodiedAct instead of CodeActGPT-5.2 · EngDesign Extendedaverage score: 49.4 -> 65.4Improved
use EmbodiedAct instead of CodeActGPT-5.2 · SciBench-107accuracy (%): 49.53 -> 54.21Improved
use EmbodiedAct instead of generative baselineQwen3-235B-Instruct · EngDesign Coreaverage score: 39.6 -> 57.0Improved
use EmbodiedAct instead of CodeActQwen3-235B-Instruct · SciBench-107accuracy (%): 55.14 -> 52.34Hurt
use EmbodiedAct instead of CodeActQwen3-8B-VL-Instruct · EngDesign Coreaverage score: 38.9 -> 40.8Mixed

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