0

ELLA: Equip Diffusion Models with LLM for Enhanced Semantic Alignment

ELLA, an Efficient Large Language Model Adapter, enhances text-to-image diffusion models by integrating powerful Large Language Models through a Timestep-Aware Semantic Connector, improving dense prompt comprehension and generation quality.

Year
2024
Venue
arXiv 2024
Authors
6
Hosting
Abstract onlyARXIV-DEFAULT

Cite

Notes

Only stored in your browser.

Attribution

Abstract & full text
arxiv.org/abs/2403.05135ARXIV-DEFAULT
TL;DR
Semantic Scholar
Attribution policy →

Abstract

Diffusion models have demonstrated remarkable performance in the domain of text-to-image generation. However, most widely used models still employ CLIP as their text encoder, which constrains their ability to comprehend dense prompts, encompassing multiple objects, detailed attributes, complex relationships, long-text alignment, etc. In this paper, we introduce an Efficient Large Language Model Adapter, termed ELLA, which equips text-to-image diffusion models with powerful Large Language Models (LLM) to enhance text alignment without training of either U-Net or LLM. To seamlessly bridge two pre-trained models, we investigate a range of semantic alignment connector designs and propose a novel module, the Timestep-Aware Semantic Connector (TSC), which dynamically extracts timestep-dependent conditions from LLM. Our approach adapts semantic features at different stages of the denoising process, assisting diffusion models in interpreting lengthy and intricate prompts over sampling timesteps. Additionally, ELLA can be readily incorporated with community models and tools to improve their prompt-following capabilities. To assess text-to-image models in dense prompt following, we introduce Dense Prompt Graph Benchmark (DPG-Bench), a challenging benchmark consisting of 1K dense prompts. Extensive experiments demonstrate the superiority of ELLA in dense prompt following compared to state-of-the-art methods, particularly in multiple object compositions involving diverse attributes and relationships.

Authors

6