TR2-D2:树搜索导向轨迹意识微调为离散的扩散微调

Sophia Tang1, Yuchen Zhu2, Molei Tao2

  • 1Department of Computer and Information Science, University of Pennsylvania.

ArXiv
|November 19, 2025
PubMed
概括

我们介绍了树搜索引导轨迹意识微调离散扩散 (TR2-D2) 的新方法,用于优化扩散模型. TR2-D2通过使用蒙特卡罗树搜索在离散序列中增强了奖励引导的轨迹生成.