Research Notes

Experiments in AI interpretability and introspection, replicating and extending research on open-source models.