| title | FewStepRL: Fine-tuning Few-step Diffusion Models with Reinforcement Learning |
|---|---|
| authors | Haitao Lin, Minsi Ren, Yang Li, Xingtong Ge, Zheyuan Hu, Peiyan Hu, Tailin Wu |
| date | 2026-11-05 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2611.08240 |
| affiliation | Tencent Hunyuan |
| upvotes | 9 (Hugging Face Daily Papers, 2026-11-12) |
| links | abs pdf hf code |
| cite | |
| related | AuK Technical Report: An Open-Source Foundational Model for Speech Generation and Editing Hy-Nomas: Towards Free-form Interactive World Generation Harness Handbook: Making Evolving Agent Harnesses Readable,Navigable, and Editable Hierarchical Sparse Attention Done Right: Toward Infinite Context Modeling Recursive Synthesis for Long-Horizon Terminal Tasks |
| abstract | Reinforcement learning (RL) has become a standard post-training method for diffusion models, but most RL objectives roll out the current policy at every iteration, and with a many-step sampler these rollouts typically dominate the cost of training. Few-step models cut this cost, yet they model an average velocity over an interval, whereas existing RL objectives are built for an instantaneous velocity with a many-step sampler. We adapt RL objectives to few-step models systematically in three phases, namely a transformation, a decomposition and an adaptation. The transformation to x-space expresses each objective we consider through a common policy prediction, at the clean end by default and at an explicitly marked level for transition or flow-map objectives, so that objectives stated over densities, velocities or rewards become comparable. The decomposition writes the gradient of each of these objectives as the sum of a self-regression toward a reference and a reward augmentation, so that a method reduces, at the gradient level, to a choice of anchors and weights. The adaptation computes each anchor from a few-step model at the end of its transport or along its tangent, with the instantaneous velocity recovered from the network itself or from an auxiliary network. This yields few-step versions of FlowGRPO, AWM, DiffusionNFT, DiffusionOPSD and ReFL in one modular codebase. In our experiments, reward-gradient-based methods are the most effective when the reward is differentiable and its gradient is affordable; otherwise DiffusionNFT-style methods generally reach high rewards in relatively few iterations despite a sometimes higher per-iteration cost. Extra inference steps show no systematic aggregate benefit in our experiments, although selected outputs visibly improve with two to four times as many sampling steps as in training. |
bibtex
@misc{lin2026fewsteprl,
title={FewStepRL: Fine-tuning Few-step Diffusion Models with Reinforcement Learning},
author={Haitao Lin and Minsi Ren and Yang Li and Xingtong Ge and Zheyuan Hu and Peiyan Hu and Tailin Wu},
year={2026},
eprint={2611.08240},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2611.08240}
}