Building and evaluating model diffing agents — AI Alignment Forum
alignmentforum.org · saved by 1 readers
This is the second in a series of informal research updates from the Google DeepMind Language Model Interpretability team, in interpretability and ad…