ReDi: Efficient Learning-Free Diffusion Inference via Trajectory Retrieval

TL;DR

ReDi is a learning-free retrieval-based framework that accelerates diffusion model inference by retrieving and continuing from similar trajectories, doubling inference speed while maintaining quality.

Contribution

ReDi introduces a novel retrieval-based approach to significantly speed up diffusion inference without additional training.

Findings

2x inference speedup demonstrated

Maintains generation quality with retrieval-based method

Effective in zero-shot cross-domain image generation

Abstract

Diffusion models show promising generation capability for a variety of data. Despite their high generation quality, the inference for diffusion models is still time-consuming due to the numerous sampling iterations required. To accelerate the inference, we propose ReDi, a simple yet learning-free Retrieval-based Diffusion sampling framework. From a precomputed knowledge base, ReDi retrieves a trajectory similar to the partially generated trajectory at an early stage of generation, skips a large portion of intermediate steps, and continues sampling from a later step in the retrieved trajectory. We theoretically prove that the generation performance of ReDi is guaranteed. Our experiments demonstrate that ReDi improves the model inference efficiency by 2x speedup. Furthermore, ReDi is able to generalize well in zero-shot cross-domain image generation such as image stylization.