Skip to yearly menu bar Skip to main content

Workshop: I Can’t Believe It’s Not Better (ICBINB): Failure Modes in the Age of Foundation Models

Pre-trained Language Models Do Not Help Auto-regressive Text-to-Image Generation

Yuhui Zhang · Brandon McKinzie · Zhe Gan · Vaishaal Shankar · Alexander Toshev


Recent advances in image tokenizers, such as VQ-VAE, have enabled text-to-image generation using auto-regressive methods, similar to language modeling. However, these methods have yet to leverage pre-trained language models, despite their adaptability to various downstream tasks. In this work, we explore this gap by adapting a pre-trained language model for auto-regressive text-to-image generation, and find that pre-trained language models offer limited help. We provide a two-fold explanation by analyzing tokens from each modality. First, we demonstrate that image tokens possess significantly different semantics compared to text tokens, rendering pre-trained language models no more effective in modeling them than randomly initialized ones. Second, the text tokens in the image-text datasets are too simple compared to normal language model pre-training data, which causes the catastrophic degradation of language models' capability.

Chat is not available.