AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

HiCache: A Plug-in Scaled-Hermite Upgrade for Taylor-Style Cache-then-Forecast Diffusion Acceleration

Liang Feng, Shikang Zheng, Jiacheng Liu, Yuqi Lin, Qinming Zhou, Peiliang Cai, Xinyu Wang, Junjie Chen, Chang Zou, Yue Ma, Linfeng Zhang

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Training-free diffusion acceleration

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace Taylor basis with scaled Hermite basisTaylorSeer on FLUX.1-dev · text-to-image N=7, O=2ImageReward: 0.9572 -> 0.9979Improved
replace Taylor predictor with Hermite predictor in ClusCaClusCa on FLUX.1-dev · N=7ImageReward: 0.9480 -> 0.9840Improved
replace Taylor predictor with Hermite predictor in ClusCaClusCa on FLUX.1-dev · N=9ImageReward: 0.8440 -> 0.8860Improved
replace Taylor basis with Hermite basisTaylorSeer on HunyuanVideo · VBench N=7, O=2VBench (%): 79.28 -> 79.65Improved
replace Taylor basis with Hermite basisTaylorSeer on HunyuanVideo · VBench N=6, O=1VBench (%): 79.78 -> 79.89Improved
replace Taylor basis with Hermite basisTaylorSeer on DiT-XL/2 · ImageNet ~7.1xFID: 4.58 -> 4.31Improved
add contraction factor sigma=0.5 to Hermite basis (vs sigma=1.0) (ablation)HiCache · FLUX.1-dev 5.55xImageReward: 0.7586 -> 0.9979Improved
vary contraction sigma among 0.4, 0.5, 0.7 (ablation)HiCache · FLUX.1-dev 5.55xImageReward: 0.9683 / 0.9979 / 0.9623Mixed
apply dual scaling alone to TaylorSeer (Hi-Taylor) (ablation)TaylorSeer · FLUX.1-dev 5.55xImageReward: 0.9572 -> 0.9624Improved
add Hermite basis on top of dual scaling (HiCache vs Hi-Taylor) (ablation)HiCache · FLUX.1-dev 5.55xImageReward: 0.9624 -> 0.9979Improved
accelerate with HiCache on image super-resolutionInf-DiT · NTIRE 2025 protocol, ~5.93x theoretical speedupPSNR: 30.87 -> 30.18Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy