Chatgpt 免費 版

We propose Auto Self-Correct ChatGPT (ASC-GPT) to solve the problem to let human teach ChatGPT to refresh/expand ChatGPT’s knowledge automatically. The core idea of our method is: Correcting/Teaching the ChatGPT’s knowledge equals to editing the related data in the whole train-dataset of ChatGPT’s policy model. We propose the procedures to implement our idea and discuss the reasons why we design these procedures.