Deep Parameter Interpolation for Scalar Conditioning

TL;DR

Deep Parameter Interpolation (DPI) introduces a versatile method to incorporate scalar inputs into neural networks by interpolating between learnable parameter sets, improving generative model performance without restricting architecture design.

Contribution

DPI provides a simple, architecture-agnostic approach for scalar conditioning in neural networks through dynamic parameter interpolation, enhancing generative models.

Findings

Improves denoising performance in diffusion models

Enhances sample quality in flow matching models

Maintains computational efficiency comparable to existing methods

Abstract

We propose deep parameter interpolation (DPI), a general-purpose method for transforming an existing deep neural network architecture into one that accepts an additional scalar input. Recent deep generative models, including diffusion models and flow matching, employ a single neural network to learn a time- or noise level-dependent vector field. Designing a network architecture to accurately represent this vector field is challenging because the network must integrate information from two different sources: a high-dimensional vector (usually an image) and a scalar. Common approaches either encode the scalar as an additional image input or combine scalar and vector information in specific network components, which restricts architecture choices. Instead, we propose to maintain two learnable parameter sets within a single network and to introduce the scalar dependency by dynamically interpolating between the parameter sets based on the scalar value during training and sampling. DPI is a simple, architecture-agnostic method for adding scalar dependence to a neural network. We demonstrate that our method improves denoising performance and enhances sample quality for both diffusion and flow matching models, while achieving computational efficiency comparable to standard scalar conditioning techniques. Code is available at https://github.com/wustl-cig/parameter_interpolation.