核心信息
Hugging Face 研究员 Sayak Paul 在 ECCV 上展示了 Flash-BoN，该项目通过将优化指标从函数评估次数（NFE）改为墙钟时间，探索扩散模型的推理时扩展。
要点
- Flash-BoN 于 9 月 11 日在 ECCV 2026 上展示，想法源自 ICCV 2025 夏威夷期间的一次交流。
- 研究主张用墙钟时间等更有界的指标来分配扩散模型的推理时计算资源。
- 这体现了推理时扩展正朝着更实用、更高效的方向发展。
Flash-BoN：重新思考扩散模型的推理时扩展
原始推文
RT @RisingSayak: We presented our work Flash-BoN on the 11th Sept at #ECCV26. It was a fulfilling experience, to say the least! @RawalRuchit told me about the idea in Hawai'i during ICCV'25, and I was immediately like, let's go! The origins of the work started with a curiosity: Change the metric for inference-time scaling algos in diffusion from number of function evals (NFE) to something more bounded, like wall-clock time. We decided to spend that (inference-time) precious compute exploring more candidates WITHOUT busting the tanks. It's so cool that Flash-BoN works across the board: T2I, T2V; different model scales; complements other techniques like BFS, ReflectionFlow; and even improves the convergence of Flow-GRPO. If you haven't checked it out yet, here's the link: https://t.co/CRsT84cJY0