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..
How to build a consistency model: Learning flow maps via self-distillation
Authors: Nicholas Boffi, Michael Albergo, Eric Vanden-Eijnden
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 5 Numerical experiments We test LSD, ESD, PSD-U, and PSD-M on the low-dimensional checkerboard dataset, as well as in the high-dimensional setting of unconditional image generation on CIFAR-10, Celeb A-64, and AFHQ-64. In each case, we study performance at fixed training time to obtain a fair comparison. ... Table 1: Benchmark results. Performance across sampling step counts for the low-dimensional checker dataset (KL divergence) and natural image datasets (FID). |
| Researcher Affiliation | Academia | Nicholas M. Boffi Carnegie Mellon University Michael S. Albergo Harvard University Eric Vanden-Eijnden Courant Institute of Mathematical Sciences |
| Pseudocode | Yes | Algorithm 1: Learning flow maps via self-distillation input: Dataset D; interpolant coefficients αt, βt; batch size M; diagonal fraction η; distillation method LD {LLSD, LESD, LPSD}. |
| Open Source Code | Yes | Associated code is available at https://github.com/nmboffi/flow-maps. |
| Open Datasets | Yes | We test LSD, ESD, PSD-U, and PSD-M on the low-dimensional checkerboard dataset, as well as in the high-dimensional setting of unconditional image generation on CIFAR-10, Celeb A-64, and AFHQ-64. ... Celeb A-64 (Liu et al., 2015) ... AFHQ-64 (Choi et al., 2020). |
| Dataset Splits | No | We test LSD, ESD, PSD-U, and PSD-M on the low-dimensional checkerboard dataset, as well as in the high-dimensional setting of unconditional image generation on CIFAR-10, Celeb A-64, and AFHQ-64. |
| Hardware Specification | Yes | Each experiment was run on a single 40GB A100 GPU. |
| Software Dependencies | No | Yes, we plan to release a well-documented open source release with the camera-ready version. Included in this will be an open-source jax implementation of the EDM2 neural network. |
| Experiment Setup | Yes | We train for 150, 000 steps with a batch size of 100, 000 and a learning rate of 10 3 with square root decay after 35, 000 steps. We use a diagonal fraction of η = 0.75, allocating 75% of each batch to the flow matching loss Lb and 25% to the self-distillation loss. The network architecture consists of a 4-layer MLP with 512 neurons per hidden layer and GELU activation functions. |