On the Closed-Form of Flow Matching: Generalization Does Not Arise from Target Stochasticity

NeurIPSOral2025

Authors: Quentin Bertrand, Anne Gagneux, Mathurin Massias, Rémi Emonet
Affiliation: INRIA
Venue: NeurIPS 2025
Track: Oral

TL;DR

We leverage the closed-form formulation of flow matching to understand its generalization…

Opening excerpt from the authors’ abstract. source

Read the paper

OpenReview

Topics

generalization flow matching rag

← All NeurIPS 2025 Oral papers · Browse the whole archive