AI Paper Atlas

Research map / Reinforcement learning / MDPs and planning

EMFuse: Energy-based Model Fusion for Decision Making

Kejie He, Yi-Chen Li, Yang Yu

ICLR 2026 · Reinforcement learning › MDPs and planning · approach: Average-reward and regularized MDPs

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
fuse by energy sum instead of parameter averaging (Soup)expert LLMs · OpenCompass subject-mixaccuracy: 60.88 -> 63.49Improved
fuse by energy sum instead of PackLLM logit packingexpert LLMs · OpenCompass subject-mixaccuracy: 63.15 -> 63.49No clear effect
energy-sum fusion instead of PackLLMexpert LLMs · OpenCompass finance-suiteaccuracy: 88.27 -> 89.21Improved
add EMSelect per-step expert selection on top of EMFuse (ablation)EMFuse · OpenCompass subject-mix and finance-suiteaccuracy: 63.49 -> 64.80; 89.21 -> 90.39Improved
add EMSelect on top of EMFuse for math sets (ablation)EMFuse · MGSM (math)accuracy: -2.40Hurt
use temporal-consistency uncertainty penalty instead of zero penalty (ablation)ADETM · D4RL HalfCheetah mediumIQM normalized return: 7.94 -> 41.83Improved
use temporal-consistency uncertainty instead of action-noise sensitivity (ablation)ADETM · D4RL Hopper mediumIQM normalized return: 37.22 -> 49.03Improved
use action-noise local sensitivity penalty instead of no penalty (ablation)ADETM · D4RL HalfCheetah mediumIQM normalized return: 7.94 -> 1.72Hurt
single ADETM instead of ensemble of 5 dynamics models (ablation)dynamics model · D4RL medium (Hopper)epoch FLOPs (T): 44.35 -> 12.32Improved
use entropy-based reweighting instead of uniform fusion weights (ablation)EMFuse · OpenCompassaccuracyNo clear effect
energy fusion of 1B experts instead of an 8B modelEMFuse · AlpacaEval domainstoken-level KL to 1B expert: 0.5063 -> 0.0391 (harmless)Improved
use EMFuse instead of Mixed-data oracle on MuJoCo mediumEMFuse · D4RL MuJoCo medium averageIQM normalized return: 47.8 -> 50.1No clear effect

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