Skip to content

๐Ÿ“– Summarizing Books with GPT-3.5


This project demostration is for educational purpose only.

๐Ÿ’ก About

This is a continuous project inspired by OpenAI's demonstration on Summarizing books with human feedback.

๐Ÿง  Rationale

Lifes are too short for worth reading books

I have a long list of classics books that I wish to read, but never get adequate time for them. While we could read the others' (or ChatGPT's) summaries, their quality are often too variated. Why don't we instruct our GPT-3.5 model to read and summarize them for me?

Forget about the reinforcement learning with human feedback et al. black magics, I simply find the concept of first summarizing small sections of a book, then summarizing those summaries into a higher-level summary, and so on is quite useful to get quality summaries. And I aim to try the method on a small dataset, and if the results are good, perhaps we could open-source our summaries, or make some color books with other generative AIs, who knows?

๐Ÿ“– Book List

  1. Artificial Intelligence: A Very Short Introduction (2018)
  2. Letters to his Son, by The Earl of Chesterfield (Coming soon)