OpenAI's Automated Interpretability from paper "Language models can explain neurons in language models". Modified by Johnny Lin to add new models/context windows.
strings that look like encoded or technical data—mixed-case alphanumeric fragments and symbol runs typical of keys, hashes, base64 blocks, or code headers.
