Stylus: Automatic Adapter Selection for Diffusion Models

TL;DR

Stylus is a method that automatically selects and composes task-specific adapters for diffusion models based on prompt keywords, improving image generation quality and efficiency.

Contribution

It introduces a three-stage approach for adapter selection and composition, along with a new dataset, StylusDocs, for evaluating adapter matching in diffusion models.

Findings

Achieves higher CLIP-FID Pareto efficiency on stable diffusion checkpoints.

Twice as preferred by humans and multimodal evaluators over the base model.

Effectively matches prompts to relevant adapters for improved image synthesis.

Abstract

Beyond scaling base models with more data or parameters, fine-tuned adapters provide an alternative way to generate high fidelity, custom images at reduced costs. As such, adapters have been widely adopted by open-source communities, accumulating a database of over 100K adapters-most of which are highly customized with insufficient descriptions. This paper explores the problem of matching the prompt to a set of relevant adapters, built on recent work that highlight the performance gains of composing adapters. We introduce Stylus, which efficiently selects and automatically composes task-specific adapters based on a prompt's keywords. Stylus outlines a three-stage approach that first summarizes adapters with improved descriptions and embeddings, retrieves relevant adapters, and then further assembles adapters based on prompts' keywords by checking how well they fit the prompt. To evaluate Stylus, we developed StylusDocs, a curated dataset featuring 75K adapters with pre-computed adapter embeddings. In our evaluation on popular Stable Diffusion checkpoints, Stylus achieves greater CLIP-FID Pareto efficiency and is twice as preferred, with humans and multimodal models as evaluators, over the base model. See stylus-diffusion.github.io for more.