Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..

Shortcutting Pre-trained Flow Matching Diffusion Models is Almost Free Lunch

Authors: Xu Cai, Yang Wu, Qianli Chen, Haoran Wu, Lichuan Xiang, Hongkai Wen

NeurIPS 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we present empirical studies to evaluate SCFM. We begin by showcasing numerical and visual comparisons, followed by comprehensive ablations in Section 5.4 (including few-shot learning, detailed in Appendix F) that explores how to achieve the most efficient and performant SCFM configuration.
Researcher Affiliation Collaboration AI Research Center, i Human Inc.. Department of Computer Science, University of Warwick. Corresponding author: EMAIL.
Pseudocode Yes The vanilla training algorithm is provided in Algorithm 1, located in Appendix A.
Open Source Code Yes Project page: shortcutfm.github.io.
Open Datasets Yes We conduct our experiments on a filtered subset of the LAION dataset Schuhmann et al. [2022], specifically the LAION-POP dataset Laion Pop [2024], which contains 600k* samples in total with aesthetic scores >0.5.
Dataset Splits Yes We conduct our experiments on a filtered subset of the LAION dataset Schuhmann et al. [2022], specifically the LAION-POP dataset Laion Pop [2024]... For evaluation, we adopt the widely used COCO-30k validation set Lin et al. [2014].
Hardware Specification Yes All training and evaluation are performed on a single NVIDIA A100 80GB GPU.
Software Dependencies No The paper mentions 'Adam W optimizer Loshchilov and Hutter [2019]' and 'Lo RA) Hu et al. [2022]' as techniques/libraries used, but does not specify version numbers for any software dependencies like programming languages, specific library versions, or frameworks (e.g., PyTorch, TensorFlow, CUDA).
Experiment Setup Yes We use the Adam W optimizer Loshchilov and Hutter [2019] with a learning rate of 2e 5, a batchsize of N =16, and set k N =0.4 in (13), chosen bootstrappingly.