5 papers · 1 filter
Meta Flow Maps enable scalable reward alignment
Peter Potaptchik, Adhi Saravanan, Abbas Mammadov +3
Controlling generative models is computationally expensive. This is because optimal alignment with a reward function--whether via inference-time steering or fine-tuning--requires e…
Itô maps for any-step SDEs
Zhengkai Pan, Peter Potaptchik, Wenxi Yao +2
Recent one-step generative models accelerate sampling by learning deterministic flow maps of the underlying dynamics. These methods rely on learning from ordinary differential equa…
A unified perspective on fine-tuning and sampling with diffusion and flow models
Carles Domingo-Enrich, Yuanqi Du, Michael S. Albergo
We study the problem of training diffusion and flow generative models to sample from target distributions defined by an exponential tilting of a base density; a formulation that su…
Discrete Flow Maps
Peter Potaptchik, Jason Yim, Adhi Saravanan +3
The sequential nature of autoregressive next-token prediction imposes a fundamental speed limit on large language models. While continuous flow models offer a path to parallel gene…
Tilt Matching for Scalable Sampling and Fine-Tuning
Peter Potaptchik, Cheuk-Kit Lee, Michael S. Albergo
We propose a simple, scalable algorithm for using stochastic interpolants to sample from unnormalized densities and for fine-tuning generative models. The approach, Tilt Matching,…