ParaUni：以强化学习驱动的层级并行信息交互增强统一多模态模型的生成能力
ParaUni: Enhance Generation in Unified Multimodal Model with Reinforcement-driven Hierarchical Parallel Information Interaction
把 VLM 各层特征并行抽取后经 Layer Integration Module 融合，作为扩散生成的条件信号，并用 RL 的层级动态调整机制对不同层施加不同奖励；做 VLM-conditioned 生成或想把多层语义注入扩散骨干的工作可直接复用这套并行条件注入与分层奖励设计。
原文依据 visual features from all VLM's layers are fed in parallel into a Layer Integration Module (LIM)
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论统一多模态模型通过将视觉语言模型（VLM）与扩散模型相结合，显著提升了视觉生成能力。然而，由于表征差异巨大，现有方法难以在充分交互与灵活实现之间取得良好平衡。考虑到 VLM 各层中包含从低层细节到高层语义的丰富且具有层级性的信息，我们提出 ParaUni。它以并行方式从 VLM 的不同层提取特征，以实现全面的信息交互，并保留灵活的分离式架构，从而增强统一多模态模型的生成能力。具体而言，来自 VLM 所有层的视觉特征被并行送入层集成模块（LIM），该模块高效地整合细粒度细节与语义抽象，并将融合后的表征作为条件提供给扩散模型。为进一步提升性能，我们揭示出这些层级对强化学习（RL）中不同奖励的响应并不相同。关键在于，我们设计了层级动态调整机制（LDAM），借助 RL 对齐这些层的层级特性，从而促进多项奖励的共同提升。大量实验表明，ParaUni 利用互补的多层特征显著提升了生成质量，并在 RL 阶段展现出推动多项奖励提升的强大潜力。
英文原文摘要 1396 字符
Unified multimodal models significantly improve visual generation by combining vision-language models (VLMs) with diffusion models. However, existing methods struggle to fully balance sufficient interaction and flexible implementation due to vast representation difference. Considering abundant and hierarchical information in VLM's layers from low-level details to high-level semantics, we propose ParaUni. It extracts features from variants VLM's layers in a Parallel way for comprehensive information interaction and retains a flexible separation architecture to enhance generation in Unified multimodal model. Concretely, visual features from all VLM's layers are fed in parallel into a Layer Integration Module (LIM), which efficiently integrates fine-grained details and semantic abstractions and provides the fused representation as a condition to the diffusion model. To further enhance performance, we reveal that these hierarchical layers respond unequally to different rewards in Reinforcement Learning (RL). Crucially, we design a Layer-wise Dynamic Adjustment Mechanism (LDAM) to facilitate multiple reward improvements that aligns the hierarchical properties of these layers using RL. Extensive experiments show ParaUni leverages complementary multi-layer features to substantially improve generation quality and shows strong potential for multiple reward advances during RL stages.