AI Paper Atlas

Research map / Reinforcement learning / Offline RL

Training-Free Adaptation of Diffusion Models via Doob's $h$-Transform

Qijie Zhu, Zeqi Ye, Han Liu, Zhaoran Wang, Minshuo Chen

ICML 2026 · Reinforcement learning › Offline RL · approach: Diffusion models for offline decision making · 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
apply DOIT-Practical correction to diffusion samplingStable Diffusion v1.5 · LAION aesthetic scoremean aesthetic score: 6.372 -> 6.726Improved
replace full-rollout DOIT-Proto with one-step surrogate (ablation)DOIT · LAION aesthetic scoremean aesthetic score: 6.714 -> 6.726No clear effect
replace full-rollout DOIT-Proto with one-step surrogate (ablation)DOIT · LAION aesthetic scoreruntime per image (s): 39.584 -> 1.712Improved
combine DOIT-Practical with Best-of-KStable Diffusion v1.5 · ImageReward K=4ImageReward: 0.702 -> 0.875Improved
combine DOIT-Practical with Best-of-KStable Diffusion v1.5 · ImageReward K=8ImageReward: 0.896 -> 1.001Improved
combine DOIT-Practical with BFSStable Diffusion v1.5 · ImageReward K=4ImageReward: 0.882 -> 0.950Improved
combine DOIT-Practical with BFSStable Diffusion v1.5 · ImageReward K=8ImageReward: 1.087 -> 1.115Improved
use DOIT-Practical instead of TTS as inference-time methodD4RL diffusion policies · D4RL locomotion averagenormalized score: 86.1 -> 87.6Improved
use DOIT-Practical instead of TTSD4RL diffusion policies · Medium-Replay HalfCheetahnormalized score: 47.8 -> 46.9Hurt
use DOIT-Practical instead of TTSD4RL diffusion policies · Medium-Replay Walker2dnormalized score: 79.3 -> 82.0Improved
lower temperature tau requires lower correction strength gamma (ablation)DOIT-Practical · LAION aesthetic scoreaesthetic score stabilityMixed

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