AI Paper Atlas

Research map / Reinforcement learning / Policy gradient and actor-critic

Dichotomous Diffusion Policy Optimization

Ruiming Liang, Yinan Zheng, Kexin ZHENG, Tianyi Tan, Jianxiong Li, Liyuan Mao, Zhihao Wang, Guang Chen, Hangjun Ye, Jingjing Liu, Jinqiao Wang, Xianyuan Zhan

ICLR 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Regularized scalable deep value learning

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
fine-tune with DIPOLE on navtrainDP-VLA · NAVSIMPDMS: 88.3 -> 89.7Improved
fine-tune with DIPOLE on navtest without ground truthDP-VLA · NAVSIMPDMS: 88.3 -> 94.8Improved
use DIPOLE instead of DPPO for fine-tuningDP-VLA · NAVSIM navtestPDMS: 89.0 -> 94.8Improved
add rejection sampling at inference (ablation)DIPOLE · ExORL Walker standaverage return: 793 -> 953Improved
add rejection sampling at inference (ablation)DIPOLE · ExORL Jaco reach-top-rightaverage return: 84 -> 117Improved
use dichotomous greedified weights instead of CFGRL (ablation)DIPOLE w/o rs · ExORL Walker standaverage return: 782 -> 793Improved
use DIPOLE instead of FQLDIPOLE · ExORL Jaco reach-top-rightaverage return: 224 -> 117Hurt
use DIPOLE instead of IFQLDIPOLE · OGBench humanoidmaze-large-navigatesuccess rate: 11 -> 6Hurt
online fine-tuning with DIPOLEDIPOLE · OGBench humanoidmaze-mediumsuccess rate: 61 -> 97Improved
vary beta and greediness factor omega (ablation)DIPOLE · ExORLaverage returnNo 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