OpenAI releases alignment research and an open-source tool that uses GPT-4 to try to interpret the behavior of individual GPT-2 “neurons and attention heads” (Kyle Wiggers/TechCrunch) 09-05-2023

Kyle Wiggers / TechCrunch:
OpenAI releases alignment research and an open-source tool that uses GPT-4 to try to interpret the behavior of individual GPT-2 “neurons and attention heads” — It’s often said that large language models (LLMs) along the lines of OpenAI’s ChatGPT are a black box, and certainly, there’s some truth to that.