facebook/adjoint_sampling

Adjoint Sampling: Highly Scalable Diffusion Samplers via Adjoint Matching

This repository contains the models for our official code release.
Additionally, it contains the specific eSEN checkpoint that we used to train those models.

Arxiv (Accepted in ICML 2025)
Authors: Aaron Havens, Benjamin Kurt Miller, Bing Yan, Carles Domingo-Enrich, Anuroop Sriram, Brandon Wood, Daniel Levine, Bin Hu, Brandon Amos, Brian Karrer, Xiang Fu, Guan-Horng Liu, Ricky T. Q. Chen

Download models

from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="facebook/adjoint_sampling",
    local_dir="models",
    allow_patterns=[
        "am/",
        "bmam/",
        "pretrain_for_bmam/",
        "torsion/",
        "esen_spice.pt",
    ],
)