用梯度下降自动给模型特征打标签，让解释更直观。
Automated Feature Labeling with Token-Space Gradient Descent
- 在标记空间用梯度优化标签表示，以语言模型作判别器。
- 成功为动物、数字等特征生成可读的单标记标签。
- 适合关注模型可解释性的研究人员使用。
我们提出一种基于标记空间梯度下降的特征标注新方法。与以往依赖语言模型生成特征语义假设的方法不同，本方法将语言模型作为判别器，直接优化标签表示以预测特征激活。该过程被建模为多目标优化问题，兼顾预测准确性、熵最小化和语言自然性。概念验证实验表明，该方法能在多种领域（如识别动物、哺乳动物、中文文本、数字）中收敛到可解释的单标记标签。尽管当前实现仅限于单标记且适用于较简单特征，结果表明标记空间梯度下降有望成为可解释性研究的重要工具。
原文摘要 · Abstract (English)
We present a novel approach to feature labeling using gradient descent in token-space. While existing methods typically use language models to generate hypotheses about feature meanings, our method directly optimizes label representations by using a language model as a discriminator to predict feature activations. We formulate this as a multi-objective optimization problem in token-space, balancing prediction accuracy, entropy minimization, and linguistic naturalness. Our proof-of-concept experiments demonstrate successful convergence to interpretable single-token labels across diverse domains, including features for detecting animals, mammals, Chinese text, and numbers. Although our current implementation is constrained to single-token labels and relatively simple features, the results suggest that token-space gradient descent could become a valuable addition to the interpretability researcher's toolkit.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。