Framework for evaluating cross-model memory insertion and negative concept steering in LLMs via anchored latent inversion.
-
Updated
Sep 7, 2026 - Python
Framework for evaluating cross-model memory insertion and negative concept steering in LLMs via anchored latent inversion.
To associate your repository with the niodoo topic, visit your repo's landing page and select "manage topics."