Interpretability in the Wild: A Circuit for Indirect Object Identification in GPT-2 Small

AI Safety Fundamentals - Interpretability in the Wild: A Circuit for Indirect Object Identification in GPT-2 Small