AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

TreeGRPO: Tree-Advantage GRPO for Online RL Post-Training of Diffusion Models

Zheng Ding, Weirui Ye

ICLR 2026 · Reinforcement learning › RL from feedback · approach: RL fine-tuning of generative models · 20 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
replace trajectory GRPO (DanceGRPO) with tree-structured sampling and advantagesSD3.5-M RL post-training · HPDv2 (trained on HPS-v2.1)iteration time (s): 173.5 -> 72.0Improved
replace DanceGRPO with TreeGRPOSD3.5-M · HPDv2 (trained on HPS-v2.1)HPS-v2.1: 0.3556 -> 0.3735Improved
replace DanceGRPO with TreeGRPOSD3.5-M · HPDv2 (trained on HPS-v2.1)ImageReward: 1.3668 -> 1.3294Hurt
replace DanceGRPO with TreeGRPOSD3.5-M · HPDv2 (trained on HPS-v2.1)ClipScore: 0.3769 -> 0.3703Hurt
replace MixGRPO with TreeGRPOSD3.5-M · HPDv2 (trained on HPS-v2.1)HPS-v2.1 / time: 0.3649 -> 0.3735; 145.4s -> 72.0sImproved
replace DanceGRPO with TreeGRPO in multi-reward trainingSD3.5-M · HPDv2 (HPS-v2.1 + ClipScore)HPS-v2.1 / time: 0.3485 -> 0.364; 184.0s -> 79.2sImproved
increase branching from k=3 to k=4 (d=3) (ablation)TreeGRPO · HPSv2HPSv2 / time: 0.3735 -> 0.3822; 70.0s -> 126.3sMixed
reduce branching from k=3 to k=2 (d=3) (ablation)TreeGRPO · HPSv2HPSv2: 0.3735 -> 0.3271Hurt
use 2 trees per prompt instead of 1 (ablation)TreeGRPO (k=3, d=3) · HPSv2HPSv2: 0.3735 -> 0.3771No clear effect
random window parameter r=0.5 instead of 0.3 or 0.7 (ablation)TreeGRPO · HPSv2HPSv2: 0.3632 / 0.3576 -> 0.3735Improved
shifting window instead of random window (ablation)TreeGRPO · ClipScoreClipScore: 0.3703 -> 0.3738Mixed
use 0.8:0.2 reward weighting instead of equal weighting (ablation)TreeGRPO multi-reward · HPDv2multi-metric balanceImproved

Later papers that compared with this method

3 later accepted papers report TreeGRPO as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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