TFG-Flow: Training-free Guidance in Multimodal Generative Flow

TL;DR

TFG-Flow is a training-free guidance method for multimodal generative flow models that effectively handles both continuous and discrete data, demonstrated on molecular design tasks for drug discovery.

Contribution

It introduces a novel training-free guidance approach for multimodal generative flows, addressing high-dimensionality and discrete data challenges.

Findings

Successfully guided molecule generation with desired properties

Effective in handling both continuous and discrete data

Shows promise for drug design applications

Abstract

Given an unconditional generative model and a predictor for a target property (e.g., a classifier), the goal of training-free guidance is to generate samples with desirable target properties without additional training. As a highly efficient technique for steering generative models toward flexible outcomes, training-free guidance has gained increasing attention in diffusion models. However, existing methods only handle data in continuous spaces, while many scientific applications involve both continuous and discrete data (referred to as multimodality). Another emerging trend is the growing use of the simple and general flow matching framework in building generative foundation models, where guided generation remains under-explored. To address this, we introduce TFG-Flow, a novel training-free guidance method for multimodal generative flow. TFG-Flow addresses the curse-of-dimensionality while maintaining the property of unbiased sampling in guiding discrete variables. We validate TFG-Flow on four molecular design tasks and show that TFG-Flow has great potential in drug design by generating molecules with desired properties.