Unified multimodal models (UMMs) aim to handle perception and generation in a single model. Yet existing UMMs still rely on a frozen, separately pretrained VAE for image generation, imposing a structural bottleneck. Naively removing it introduces a quality gap, as the model must learn both high-level structure and low-level details from raw pixels. In this paper, we propose Representation Forcing (RF), a technique that closes this gap by making representation prediction a native capability of the model. Concretely, RF forces the decoder to autoregressively predict visual representations as intermediate tokens before pixels; these tokens then stay in context to guide pixel diffusion within the same backbone. By turning representations from perception outputs into generation targets, RF eliminates the need for any external generative latent space. We find that RF benefits both understanding and generation. On image generation, our pixel-space model with RF matches state-of-the-art VAE-based unified models. On image understanding, pixel-space RF generally outperforms its VAE-based variant. Together, these results offer an effective step toward end-to-end, bottleneck-free UMMs.

翻译：统一多模态模型（UMMs）旨在单个模型中处理感知与生成任务。然而现有UMMs仍依赖冻结且单独预训练的VAE进行图像生成，这引入了结构性瓶颈。简单移除该模块会导致质量下降，因为模型必须从原始像素中同时学习高层结构与低层细节。本文提出表示强迫（RF）技术，通过使表示预测成为模型原生能力来消除这一质量差距。具体而言，RF强制解码器以自回归方式将视觉表示作为中间令牌预测，随后将这些令牌保留在上下文中，在同一骨干网络内引导像素扩散。通过将表示从感知输出转化为生成目标，RF消除了对外部生成隐空间的依赖。我们发现RF对理解与生成任务均有裨益。在图像生成方面，采用RF的像素空间模型性能与基于VAE的最优统一模型持平。在图像理解方面，基于像素空间的RF普遍优于其VAE变体。这些结果共同为构建无瓶颈的端到端UMMs提供了有效途径。