J-lens preferentially reads from coordinate-aligned subspaces

J-lens is a method designed to selectively read information from coordinate-aligned subspaces in data representations, enhancing interpretability and control in AI models.