Earlier this year, I shared a video (see here) that Andrej Karpathy made about how he uses large language models (LLMs). According to his website, Karpathy was “a research scientist and founder member at OpenAI” before spending several years at Tesla as Senior Director of AI. His video jumped out at me because he uses LLMs to help him read and understand books and research papers. He did this manually, and he described how painful it was, which I summarized in this post.

He’s taken his habit of reading with LLMs even further and developed a process with three passes: first, manual reading; second, having the LLM “explain and summarize”; and third, Q&A with the LLM. See his post about this here.

In addition, he solved his own pain: He shared that he’s built a simple tool for implementing this process with less friction. It enables him to read books in ePub format and feed the text of the book to the LLM, so he can ask the LLM questions while he reads. See more details in his post here.

He shared the code for his tool so others can use it, which is pretty cool. It’s available on GitHub here.

This sounds really interesting, and I can’t wait to give it a try.