Compositional Sculpting of Iterative Generative Processes

TL;DR

This paper introduces Compositional Sculpting, a method for combining iterative generative models like GFlowNets and diffusion models, enabling flexible composition and improved sample generation for images and molecules.

Contribution

It presents a general framework for compositional modeling of iterative generative processes and introduces classifier-guided sampling techniques for these compositions.

Findings

Effective compositional methods for GFlowNets and diffusion models.

Successful application to image and molecular generation.

Novel binary operations for combining distributions.

Abstract

High training costs of generative models and the need to fine-tune them for specific tasks have created a strong interest in model reuse and composition. A key challenge in composing iterative generative processes, such as GFlowNets and diffusion models, is that to realize the desired target distribution, all steps of the generative process need to be coordinated, and satisfy delicate balance conditions. In this work, we propose Compositional Sculpting: a general approach for defining compositions of iterative generative processes. We then introduce a method for sampling from these compositions built on classifier guidance. We showcase ways to accomplish compositional sculpting in both GFlowNets and diffusion models. We highlight two binary operations $\unicode{x2014}$ the harmonic mean ($p_1 \otimes p_2$) and the contrast ($p_1 \unicode{x25D1}\,p_2$) between pairs, and the generalization of these operations to multiple component distributions. We offer empirical results on image and molecular generation tasks.