arXiv 2506.05350
Contrastive Flow Matching
By George Stoica, Vivek Ramanujan, et al.
Published 2025-06-05
Citation lineage
Review the prior work and downstream research connected to this paper.
Unconditional flow-matching trains diffusion models to transport samples from a source distribution to a target distribution by enforcing that the flows between sample pairs are unique. However, in conditional settings (e.g., class-conditioned models), this uniqueness is no longer guaranteed--flows from different conditions may overlap, leading to more ambiguous generations. We introduce Contrastive Flow Matching, a…