Research

DiffusionART: Efficient, Stable and Scalable Reinforcement of Diffusion Models

Single source1 source8 Oct 2026

Online reinforcement learning (RL) fine-tunes diffusion models with black-box rewards, but reverse-process methods such as Flow-GRPO are limited in efficiency by poor credit assignment, while forward-process methods such as DiffusionNFT are efficient yet unstable without carefully tuned exponential-moving-average (EMA) off-policy training.

Reported by 1 source

arxiv.orghttps://arxiv.org/abs/2610.12572

← Back to all developments · Signals are a development reported by more than one source. See the methodology.

Research

DiffusionART: Efficient, Stable and Scalable Reinforcement of Diffusion Models

Single source1 source8 Oct 2026

Reported by 1 source

arxiv.orghttps://arxiv.org/abs/2610.12572

← Back to all developments · Signals are a development reported by more than one source. See the methodology.