Interpretability

Prompt 調整遇到邊界？本文探討如何透過稀疏自編碼器（SAE），從模型內部拆解出獨立特徵，並直接調整特徵活化值，實現比 Prompt Engineering 更精確的模型行為操控。

Have you hit the limits of prompt tuning? This piece looks at how sparse autoencoders, or SAEs, decompose independent features from inside a model and directly adjust feature activations, enabling more precise control of model behavior than prompt engineering.