Unified Autoregressive Visual Generation and Understanding with Continuous Tokens
This work addresses the challenge of integrating image generation and understanding in a single model, which is incremental as it builds on existing autoregressive and pre-trained LLM methods.
The authors tackled the problem of joint visual generation and understanding by proposing UniFluid, a unified autoregressive framework using continuous tokens, achieving results comparable to or exceeding single-task baselines on both tasks.
We present UniFluid, a unified autoregressive framework for joint visual generation and understanding leveraging continuous visual tokens. Our unified autoregressive architecture processes multimodal image and text inputs, generating discrete tokens for text and continuous tokens for image. We find though there is an inherent trade-off between the image generation and understanding task, a carefully tuned training recipe enables them to improve each other. By selecting an appropriate loss balance weight, the unified model achieves results comparable to or exceeding those of single-task baselines on both tasks. Furthermore, we demonstrate that employing stronger pre-trained LLMs and random-order generation during training is important to achieve high-fidelity image generation within this unified framework. Built upon the Gemma model series, UniFluid exhibits competitive performance across both image generation and understanding, demonstrating strong transferability to various downstream tasks, including image editing for generation, as well as visual captioning and question answering for understanding.