研究人员在图变换器中发现了一种称为距离失配训练的现象，即模型的通信分配与其为给定任务相关信息的位置不匹配。他们开发了一个合成基准来研究这个问题，发现最佳通信距离偏差随任务局部性而变化。一个由任务距离目标驱动的自适应控制器，显著提高了相对于基线的性能，突显了任务特定控制的重要性。 AI
影响 提出通过将通信与任务相关性对齐来诊断和改进图变换器性能的新方法。
排序理由 关于图变换器训练特定方面的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
图变换器训练问题已识别，并提出自适应控制
报道来源 [2]
-
Distance-Misaligned Training in Graph Transformers and Adaptive Graph-Aware Control
arXiv:2604.22413v1 Announce Type: new Abstract: Graph Transformers can mix information globally, but this flexibility also creates failure modes: some tasks require long-range communication while others are better served by local interaction. We study this through a synthetic nod…
-
Distance-Misaligned Training in Graph Transformers and Adaptive Graph-Aware Control
Graph Transformers can mix information globally, but this flexibility also creates failure modes: some tasks require long-range communication while others are better served by local interaction. We study this through a synthetic node-classification benchmark on contextual stochas…