研究人员开发了一种新颖的方法，将Transformer语言模型中不透明的注意力机制转换为可执行的Python程序。该方法包括分析特定注意力头的注意力矩阵，然后提示预训练语言模型生成复制这些模式的代码。生成的程序可以用来替换神经网络注意力头，对模型性能的影响极小，从而促进神经网络的符号透明度。 AI
影响 增强Transformer模型的解释性和符号透明度。
排序理由 该集群包含一篇详细介绍深度学习模型新解释方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2606.19317v1 Announce Type: cross Abstract: A longstanding goal of research on interpretable deep learning is to replace opaque neural computations with human-meaningful symbolic descriptions. In this paper, we propose an approach for approximating the behavior of component…
A longstanding goal of research on interpretable deep learning is to replace opaque neural computations with human-meaningful symbolic descriptions. In this paper, we propose an approach for approximating the behavior of components of deep networks with executable programs. We fo…
A longstanding goal of research on interpretable deep learning is to replace opaque neural computations with human-meaningful symbolic descriptions. In this paper, we propose an approach for approximating the behavior of components of deep networks with executable programs. We fo…