GAS: Improving Discretization of Diffusion ODEs via Generalized Adversarial Solver

2025-10-20

AI Summary

Diffusion models achieve high-fidelity generation but suffer from substantial computational overhead due to multi-step ODE sampling. While existing gradient-based solvers reduce function evaluations, they rely on intricate training strategies and often fail to preserve fine-grained details. To address this, we propose the Generalized Adversarial Solver (GAS), which simplifies the parameterization of ODE solvers, incorporates a distillation loss, and introduces discriminator-guided adversarial training—without requiring auxiliary training techniques. GAS effectively suppresses artifacts and enhances detail recovery. Experiments demonstrate that, under comparable computational budgets, GAS achieves superior generation quality over state-of-the-art solver training methods using fewer sampling steps—particularly improving texture sharpness and structural fidelity.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

While diffusion models achieve state-of-the-art generation quality, they still suffer from computationally expensive sampling. Recent works address this issue with gradient-based optimization methods that distill a few-step ODE diffusion solver from the full sampling process, reducing the number of function evaluations from dozens to just a few. However, these approaches often rely on intricate training techniques and do not explicitly focus on preserving fine-grained details. In this paper, we introduce the Generalized Solver: a simple parameterization of the ODE sampler that does not require additional training tricks and improves quality over existing approaches. We further combine the original distillation loss with adversarial training, which mitigates artifacts and enhances detail fidelity. We call the resulting method the Generalized Adversarial Solver and demonstrate its superior performance compared to existing solver training methods under similar resource constraints. Code is available at https://github.com/3145tttt/GAS.

Problem

• Reducing computational cost of diffusion model sampling
• Improving detail preservation in distilled ODE solvers
• Mitigating artifacts through adversarial training techniques

Innovation

• Generalized Solver parameterizes ODE sampler without training tricks
• Combines distillation loss with adversarial training for detail fidelity
• Mitigates artifacts and enhances performance under resource constraints