Designing Instance-Level Sampling Schedules via REINFORCE with James-Stein Shrinkage

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yu, Peiyu, Kothawade, Suraj, Xie, Sirui, Wu, Ying Nian, Fei, Hongliang
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917435707228160
author Yu, Peiyu
Kothawade, Suraj
Xie, Sirui
Wu, Ying Nian
Fei, Hongliang
author_facet Yu, Peiyu
Kothawade, Suraj
Xie, Sirui
Wu, Ying Nian
Fei, Hongliang
contents Most post-training methods for text-to-image samplers focus on model weights: either fine-tuning the backbone for alignment or distilling it for few-step efficiency. We take a different route: rescheduling the sampling timeline of a frozen sampler. Instead of a fixed, global schedule, we learn instance-level (prompt- and noise-conditioned) schedules through a single-pass Dirichlet policy. To ensure accurate gradient estimates in high-dimensional policy learning, we introduce a novel reward baseline based on a principled James-Stein estimator; it provably achieves lower estimation errors than commonly used variants and leads to superior performance. Our rescheduled samplers consistently improve text-image alignment including text rendering and compositional control across modern Stable Diffusion and Flux model families. Additionally, a 5-step Flux-Dev sampler with our schedules can attain generation quality comparable to deliberately distilled samplers like Flux-Schnell. We thus position our scheduling framework as an emerging model-agnostic post-training lever that unlocks additional generative potential in pretrained samplers.
format Preprint
id arxiv_https___arxiv_org_abs_2511_22177
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Designing Instance-Level Sampling Schedules via REINFORCE with James-Stein Shrinkage
Yu, Peiyu
Kothawade, Suraj
Xie, Sirui
Wu, Ying Nian
Fei, Hongliang
Machine Learning
Computer Vision and Pattern Recognition
Most post-training methods for text-to-image samplers focus on model weights: either fine-tuning the backbone for alignment or distilling it for few-step efficiency. We take a different route: rescheduling the sampling timeline of a frozen sampler. Instead of a fixed, global schedule, we learn instance-level (prompt- and noise-conditioned) schedules through a single-pass Dirichlet policy. To ensure accurate gradient estimates in high-dimensional policy learning, we introduce a novel reward baseline based on a principled James-Stein estimator; it provably achieves lower estimation errors than commonly used variants and leads to superior performance. Our rescheduled samplers consistently improve text-image alignment including text rendering and compositional control across modern Stable Diffusion and Flux model families. Additionally, a 5-step Flux-Dev sampler with our schedules can attain generation quality comparable to deliberately distilled samplers like Flux-Schnell. We thus position our scheduling framework as an emerging model-agnostic post-training lever that unlocks additional generative potential in pretrained samplers.
title Designing Instance-Level Sampling Schedules via REINFORCE with James-Stein Shrinkage
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2511.22177