语义路由：探索多层LLM特征加权用于扩散变换器
Semantic Routing: Exploring Multi-Layer LLM Feature Weighting for Diffusion Transformers
AI总结：
本文提出深度-wise语义路由方法，通过多层LLM特征加权提升文本到图像生成的准确性与稳定性。
摘要：
最近基于DiT的文本到图像模型越来越多地采用LLM作为文本编码器，但文本条件仍大多静态，并且通常只使用单个LLM层，尽管在LLM层之间存在明显的语义层次结构，并且在扩散时间和网络深度上存在非平稳的去噪动态。为了更好地匹配DiT生成的动态过程，从而增强扩散模型的生成能力，我们引入了一个统一的归一化凸融合框架，配备了轻量级门控，以系统地组织多层LLM隐藏状态，通过时间-wise、深度-wise和联合融合。实验表明，深度-wise语义路由是更优的条件策略，一致地提高了文本-图像对齐和组合生成（例如，在GenAI-Bench计数任务上提高了+9.97）。相反，我们发现纯时间-wise融合可能会反常地降低视觉生成保真度。我们归因于训练-推理轨迹不匹配：在无分类器引导下，名义上的时间步长无法跟踪有效的SNR，导致推理过程中语义时间错位的特征注入。总体而言，我们的结果将深度-wise路由定位为一个强大且有效的基线，并突显了轨迹感知信号对于实现稳健的时间依赖性条件的重要性。
英文摘要：
Recent DiT-based text-to-image models increasingly adopt LLMs as text encoders, yet text conditioning remains largely static and often utilizes only a single LLM layer, despite pronounced semantic hierarchy across LLM layers and non-stationary denoising dynamics over both diffusion time and network depth. To better match the dynamic process of DiT generation and thereby enhance the diffusion model's generative capability, we introduce a unified normalized convex fusion framework equipped with lightweight gates to systematically organize multi-layer LLM hidden states via time-wise, depth-wise, and joint fusion. Experiments establish Depth-wise Semantic Routing as the superior conditioning strategy, consistently improving text-image alignment and compositional generation (e.g., +9.97 on the GenAI-Bench Counting task). Conversely, we find that purely time-wise fusion can paradoxically degrade visual generation fidelity. We attribute this to a train-inference trajectory mismatch: under classifier-free guidance, nominal timesteps fail to track the effective SNR, causing semantically mistimed feature injection during inference. Overall, our results position depth-wise routing as a strong and effective baseline and highlight the critical need for trajectory-aware signals to enable robust time-dependent conditioning.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
