[Question] Open-source AI safety projects?

Has anyone compiled a list of open-source projects?

I’m particularly interested in model evaluation (≈metr.org)