用可分解探针揭示少步扩散模型的条件响应机制，定位关键行为来源。
A Decomposable Probe for Few-Step Diffusion Models: Prompt, Latent, and Score Selectivity across Backbone Families and Distillation Paradigms
- 通过三层次扰动探测提示、隐空间和得分层的响应特性。
- 发现隐空间层能近乎二元检测修正流结构模型，仅限于特定架构。
- 探针设计统一适用于23个模型，结果可复现且具统计可信度。
少步蒸馏扩散模型将文本到图像生成从约50次网络评估压缩至1-8次，但质量差距常以单一FID/CLIP数值概括，无法判断条件响应变化的维度，也无法区分行为源于架构、蒸馏目标还是扩散模型本身。本文引入可分解探针，在三个层级（提示编码器、去噪器输入、去噪器输出）的三种模式（均值、方差、缩放）下注入六种强度的受控扰动，报告基于Inception特征的自助中位数Bures W2^2选择性比。在覆盖五类主干架构（SDXL、SD1.5、SD3.5、PixArt-alpha、FLUX）、三类结构（UNet、DiT、MMDiT）及五种蒸馏范式、共23个模型的统一估计器下，三层次读取出三个经验上可分离的因素：提示层为通用提示均值响应（基准通道，非判别器），隐空间层反映预测类型，得分层反映蒸馏目标。主要发现：隐空间层是修正流架构的近似二元检测器——其比率仅在低至中等强度范围内持续超过1，且仅对修正流模型（SD3.5、FLUX）成立；所有ε-预测模型均不满足。以ε-预测对照模型（PixArt-alpha）排除宽尺度T5条件影响，指纹在对抗性（ADD）蒸馏中仍稳定存在，无论教师或学生身份。两个次要发现：在UNet家族中，4步ADD vs 非ADD对比呈现典型差异；在UNet与DiT中，轨迹展开早期强扰动引发显著得分尖峰。所有比率均在单估计器下具有置信区间可引用性；已发布每单元表格与估计器代码。
原文摘要 · Abstract (English)
Few-step distilled diffusion students cut text-to-image inference from ~50 to 1-8 network evaluations, but the quality gap is usually summarised by a single FID/CLIP scalar that cannot say which axis of the conditioning response changed, nor whether a behaviour comes from the architecture, the distillation objective, or simply from being a diffusion model. We replace the scalar with a decomposable probe that injects controlled perturbations along three layers (prompt encoder, denoiser input, denoiser output) under three modes (mean, variance, scale) and six strengths, reporting a bootstrap-median Bures W2^2 selectivity ratio on Inception features. Under a single matched estimator across 23 models -- five teachers and 18 distilled students spanning five backbone families (SDXL, SD1.5, SD3.5, PixArt-alpha, FLUX), three architecture classes (UNet, DiT, MMDiT), and five distillation paradigms -- the three layers read three empirically separable factors: the prompt layer is a universal prompt-mean response (a sanity channel, not a discriminator), the latent layer reads the prediction type, and the score layer reads the distillation objective. Our main result: within this sweep, the latent layer is a near-binary detector of rectified-flow backbones. Its ratio exceeds 1 across a sustained low-to-mid band only for rectified-flow models (SD3.5, FLUX); no epsilon-prediction model qualifies. A matched epsilon-prediction control (PixArt-alpha) rules out wide-T5 conditioning, and the fingerprint survives adversarial (ADD) distillation as both teacher and student. Two secondary score-layer findings hold under narrower scopes: a canonical 4-step ADD-vs-rest contrast on the UNet families with a non-ADD baseline, and a CI-separated trajectory-rollout early-strength score spike on both UNet and DiT. All ratios are CI-citable under one estimator; we release the per-cell tables and the estimator.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。