This repository contains the official implementation of the following paper:
LDGen: Enhancing Text-to-Image Synthesis via Large Language Model-Driven Language Representation
LDGen integrates large language models (LLMs) into existing text-to-image models, supporting zero-shot multilingual text-to-image generation.