Bi-Anchor Interpolation Solver for Accelerating Generative Modeling

2026-01-29

AI Summary

Flow Matching models rely on iterative ODE solvers, resulting in high generation latency, while existing acceleration methods often compromise sample quality or incur substantial training costs and lack plug-and-play compatibility. This work proposes BA-solver, a novel approach that, without fine-tuning the frozen backbone network, introduces a lightweight SideNet comprising only 1–2% of the backbone’s parameters. BA-solver uniquely integrates bidirectional time-awareness with dual-anchor velocity interpolation to enable efficient high-order numerical integration. The method drastically reduces the number of neural function evaluations (NFE), achieving on ImageNet-256² the same generation quality as a conventional Euler solver with over 100 NFE using merely 10 NFE, while maintaining high fidelity even at just 5 NFE. It offers low training overhead, high-quality synthesis, and seamless plug-and-play compatibility.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Sampling/Simulation-based Search
• Natural Language Processing: Generation

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Flow Matching (FM) models have emerged as a leading paradigm for high-fidelity synthesis. However, their reliance on iterative Ordinary Differential Equation (ODE) solving creates a significant latency bottleneck. Existing solutions face a dichotomy: training-free solvers suffer from significant performance degradation at low Neural Function Evaluations (NFEs), while training-based one- or few-steps generation methods incur prohibitive training costs and lack plug-and-play versatility. To bridge this gap, we propose the Bi-Anchor Interpolation Solver (BA-solver). BA-solver retains the versatility of standard training-free solvers while achieving significant acceleration by introducing a lightweight SideNet (1-2% backbone size) alongside the frozen backbone. Specifically, our method is founded on two synergistic components: \textbf{1) Bidirectional Temporal Perception}, where the SideNet learns to approximate both future and historical velocities without retraining the heavy backbone; and 2) Bi-Anchor Velocity Integration, which utilizes the SideNet with two anchor velocities to efficiently approximate intermediate velocities for batched high-order integration. By utilizing the backbone to establish high-precision ``anchors''and the SideNet to densify the trajectory, BA-solver enables large interval sizes with minimized error. Empirical results on ImageNet-256^2 demonstrate that BA-solver achieves generation quality comparable to 100+ NFEs Euler solver in just 10 NFEs and maintains high fidelity in as few as 5 NFEs, incurring negligible training costs. Furthermore, BA-solver ensures seamless integration with existing generative pipelines, facilitating downstream tasks such as image editing.

Problem

• Flow Matching
• ODE solver
• generative modeling
• latency bottleneck
• Neural Function Evaluations

Innovation

• Flow Matching
• Bi-Anchor Interpolation
• SideNet
• Training-Free Acceleration
• ODE Solver