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..
High-Order Flow Matching: Unified Framework and Sharp Statistical Rates
Authors: Maojiang Su, Jerry Yao-Chieh Hu, Yi-Chen Lee, Ning Zhu, Jui-Hui Chung, Shang Wu, Zhao Song, Minshuo Chen, Han Liu
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | To provide empirical support for the proposed High-Order Flow Matching (HOFM) framework, we conduct a series of synthetic experiments designed to evaluate the practical benefits of incorporating higher-order dynamics. We compare the performance of standard first-order flow matching (equivalent to our framework with K = 1) against second-order flow matching (K = 2). ... Across all three target distributions and for every sampling step count (10, 50, and 100), the second-order model achieves a lower Wasserstein distance than the first-order model. |
| Researcher Affiliation | Academia | Northwestern University National Taiwan University University of Glasgow Princeton University Simon Institute of Computing, UC Berkeley |
| Pseudocode | No | The paper describes methodologies through definitions, theorems, and mathematical derivations (e.g., Definitions 3.1, 3.2, Theorems 3.1, 3.2, 3.3, 3.4, 4.1, 4.2, 4.3, 4.4) without presenting any explicit pseudocode blocks or algorithms. |
| Open Source Code | No | The paper does not provide any explicit statements about releasing code for the methodology described, nor does it include links to source code repositories. A citation to a related work titled 'Flow matching guide and code' by Lipman et al. (2024) is made, but this refers to a different paper's code, not the current one. |
| Open Datasets | Yes | We evaluate the models on 2D density matching tasks, transitioning a standard multivariate Gaussian distribution, π0, to three complex target distributions, π1. Following the experimental setting in [Chen et al., 2025], we use target distributions shaped as: (1) a square, (2) two intertwined spirals, and (3) three intertwined spirals. |
| Dataset Splits | No | The paper mentions evaluating models on '2D density matching tasks' with 'synthetic experiments' and specific target distributions (square, spirals). However, it does not specify any training, validation, or test splits for these datasets. Synthetic datasets are often generated on the fly, making explicit splits less common to report. |
| Hardware Specification | No | This research was supported in part through the computational resources and staff contributions provided for the Quest high performance computing facility at Northwestern University which is jointly supported by the Office of the Provost, the Office for Research, and Northwestern University Information Technology. |
| Software Dependencies | No | The paper does not provide specific details or version numbers for any software libraries, frameworks, or programming languages used for the implementation of their methodology or experiments. |
| Experiment Setup | No | The paper mentions 'sampling step count (10, 50, and 100)' as a parameter for evaluation in Section O.2, but does not provide specific hyperparameters for model training such as learning rates, batch sizes, optimizer types, or model-specific configuration details for the flow matching or transformer networks. |