---
title: "Optimize Your Sampling: Tuned Diffusion Sampling with Bayesian Optimization"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.18040v1"
permalink: "https://feed.astralor.com/items/itm_3c7f0c7b652383c510a06b2c"
published: "2026-08-18T17:33:00.000Z"
score: 77
channel: "ai"
tags: ["OYS","Diffusion Models","Text-to-Image Generation","Bayesian Optimization","Sampling Schedule"]
---

# Optimize Your Sampling: Tuned Diffusion Sampling with Bayesian Optimization

## AI commentary

> This paper is worth attention because it breaks away from the usual solver-improvement approach and directly black-box optimizes sampling timesteps, a simple yet surprisingly effective idea. For engineers and researchers working on diffusion inference acceleration, OYS can be plugged into existing pipelines almost for free, especially in low-latency deployment scenarios.

## At a glance

Sampling from a diffusion model typically requires many forward passes through a large neural network, making generation computationally expensive. While much work has focused on efficient solvers and samplers, comparatively little attention has been paid to selecting the sampling timesteps themselves. A recent line of work optimizes theoretically derived surrogates for sample quality rather than the quality metric itself. This paper proposes Optimizing Your Sampling (OYS), which treats timestep selection as a black-box optimization problem and optimizes the target metric directly with Bayesian optimization. OYS outperforms both default schedules and those from Align Your Steps on text-to-image generation, and improves over default schedules on inpainting and other image tasks, in both quantitative and human evaluations. OYS requires no additional training, is applicable even to distilled models, and improves both simple and sophisticated samplers such as Euler and DPM-Solver++. A 5-step OYS schedule retains 89%-94% of the quality of a 50-step schedule while reducing inference cost by 10x.

## Key points

- OYS treats timestep selection in diffusion models as a black-box optimization problem, directly optimizing the target metric without extra training.
- OYS-generated sampling schedules outperform default schedules and Align Your Steps on text-to-image, inpainting, and other image tasks.
- A 5-step OYS schedule retains 89%-94% of the quality of a 50-step schedule while reducing inference cost by 10x.
- OYS is applicable to distilled models and improves both simple and sophisticated samplers such as Euler and DPM-Solver++.

## Full text

Sampling from a diffusion model typically requires many forward passes through a large neural network, making generation computationally expensive. While much work has focused on efficient solvers and samplers, comparatively little attention has been paid to selecting the sampling timesteps themselves. A recent line of work optimizes theoretically derived surrogates for sample quality rather than the quality metric itself. We propose Optimizing Your Sampling (OYS), which instead treats timestep selection as a black-box optimization problem, optimizing the target metric directly with Bayesian optimization. OYS outperforms both the default schedules and those of Align Your Steps on text-to-image generation, and improves over the default schedules on inpainting and other image tasks, in both quantitative and human evaluations. OYS requires no additional training, is applicable even to distilled models, and improves both simple and sophisticated samplers such as Euler and DPM-Solver++. A 5-step OYS schedule retains 89%-94% of the quality of a 50-step schedule while reducing inference cost by 10x.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_3c7f0c7b652383c510a06b2c