面向基于大语言模型的运动轨迹生成与验证的自一致性方法
Self-Consistency for LLM-Based Motion Trajectory Generation and Verification
把自一致性从文本推理迁移到视觉域：用「原型轨迹 + 几何变换群」定义形状族，通过聚类一致轨迹来提升 LLM 生成运动轨迹的准确率，并可作为 VLM 轨迹验证的免训练后处理手段。
原文依据 we first sample diverse motion trajectories from an LLM, and then identify groups of consistent trajectories via clustering
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论自一致性已被证明是一种以轻量、无监督方式提升大语言模型在自然语言推理任务上性能的有效技术。在本工作中，我们研究如何将自一致性适配到视觉领域。具体而言，我们考虑由大语言模型生成的运动图形轨迹的生成与验证。给定一个提示词（例如「让圆沿螺旋路径移动」），我们首先从大语言模型中采样多样化的运动轨迹，然后通过聚类识别出一致轨迹的组。我们的关键洞见是：将某个提示词所关联的形状族建模为一条原型轨迹与一组几何变换（例如刚性变换、相似变换和仿射变换）的组合。于是，若两条轨迹可以在该变换群所允许的形变下相互转换，则它们可被视为一致的。我们提出一种算法，利用一组候选变换群之间的层次关系来自动恢复形状族。我们的方法将基于大语言模型的轨迹生成准确率提升了 4-6%。我们进一步将方法扩展到支持验证，观察到相较视觉语言模型基线有 11% 的精确率提升。我们的代码与数据集见 https://majiaju.io/trajectory-self-consistency。
英文原文摘要 1275 字符
Self-consistency has proven to be an effective technique for improving LLM performance on natural language reasoning tasks in a lightweight, unsupervised manner. In this work, we study how to adapt self-consistency to visual domains. Specifically, we consider the generation and verification of LLM-produced motion graphics trajectories. Given a prompt (e.g., "Move the circle in a spiral path"), we first sample diverse motion trajectories from an LLM, and then identify groups of consistent trajectories via clustering. Our key insight is to model the family of shapes associated with a prompt as a prototype trajectory paired with a group of geometric transformations (e.g., rigid, similarity, and affine). Two trajectories can then be considered consistent if one can be transformed into the other under the warps allowable by the transformation group. We propose an algorithm that automatically recovers a shape family, using hierarchical relationships between a set of candidate transformation groups. Our approach improves the accuracy of LLM-based trajectory generation by 4-6%. We further extend our method to support verification, observing 11% precision gains over VLM baselines. Our code and dataset are available at https://majiaju.io/trajectory-self-consistency