AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

TDM-R1: Reinforcing Few-Step Diffusion Models with Non-Differentiable Reward

Yihong Luo, Tianyang Hu, Weijian Luo, Jing Tang

ICML 2026 · Image generation › Fast diffusion sampling · approach: Accelerated diffusion policies · 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 TDM-R1 RL with GenEval rewardTDM-SD3.5-M (4 NFE) · GenEval overallscore: 0.61 -> 0.92Improved
apply TDM-R1 with OCR rewardTDM-SD3.5-M (4 NFE) · OCR accuracyscore: 0.55 -> 0.95Improved
train on GenEval rewardTDM-R1 · DrawBench ImageRewardImageReward: 0.99 -> 1.11Improved
train on OCR reward onlyTDM-R1 · GenEvalscore: 0.61 -> 0.67Improved
replace TDM-R1 surrogate-reward with direct RL loss (DGPO + TDM distillation) (ablation)TDM · OCR eval accuracyOCR accuracyHurt
use deterministic trajectories instead of stochastic sampling (ablation)TDM-R1 · OCR eval accuracyOCR accuracyImproved
use dynamic surrogate reward instead of frozen reward (ablation)TDM-R1 · OCR eval accuracyOCR accuracyImproved
use EMA of p_phi as dynamic reference instead of static reference (ablation)TDM-R1 · OCR eval accuracyOCR accuracyImproved
RL directly on few-step model rather than distilling a DGPO-finetuned teacher (ablation)TDM-R1 · OCR eval accuracyOCR accuracy: 0.91 -> 0.95Improved
4-NFE TDM-R1 generator vs 80-NFE surrogate reward diffusion p_phi (ablation)TDM-R1 · GenEvalscore: 0.89 -> 0.92Improved
apply TDM-R1 to Z-Image 6B with HPSv3 rewardZ-Image-Turbo · HPSv3HPSv3: 9.13 -> 9.90Improved
compare with DGPO/Flow-GRPO on image qualityTDM-R1 · DrawBench AestheticAesthetic: DGPO 5.31 vs TDM-R1 5.42Improved

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