Unleashing the Potential of Large Language Models for Text-to-Image Generation through Autoregressive Representation Alignment

xing xie, Jiawei Liu, Ziyue Lin, Huijie Fan, Zhi Han, Yandong Tang, Liangqiong Qu

Unleashing the Potential of Large Language Models for Text-to-Image Generation through Autoregressive Representation Alignment: 16 upvotes on Hugging Face Daily Papers, #15 of 50 papers on 2025-03-11. Day-by-day upvote history.

We present Autoregressive Representation Alignment (ARRA), a new training framework that unlocks global-coherent text-to-image generation in autoregressive LLMs without architectural changes. Unlike prior work that requires complex architectural redesigns, ARRA aligns LLM hidden states with visual representations from external visual foundational models via a global visual alignment loss and a hybrid token, <HYBNEXT>. This token enforces dual constraints: local next-token prediction and global semantic distillation, enabling LLMs to implicitly learn spatial and contextual coherence while retaining their original autoregressive paradigm. Extensive experiments validate ARRA's plug-and-play versatility. When training from text-generation-only LLMs or random initialization, ARRA reduces FID by 25.5% (MIMIC-CXR), 8.8% (DeepEyeNet), and 7.5% (ImageNet) for advanced autoregressive LLMs like Chameleon and LlamaGen, all without framework modifications. For domain adaption, ARRA aligns general-purpose LLMs with specialized models (e.g., BioMedCLIP), achieving an 18.6% FID reduction over direct fine-tuning on medical imaging (MIMIC-CXR). By demonstrating that training objective redesign -- not just architectural innovation -- can resolve cross-modal global coherence challenges, ARRA offers a complementary paradigm for advancing autoregressive models. Code and models will be released to advance autoregressive image generation.

Paper page on Hugging Face · arXiv

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.