Generating answers based on the contents of Scrapbox using GPT-3/ChatGPT
To generate responses based on the content of Scrapbox using GPT-3/ChatGPT, it is necessary to first perform a vector-based embedding search. This step is not mandatory, but it can be considered as a form of tuning or optimization tkgshn.icon. The flexibility in handling questions seems to be quite high. When a user throws a prompt, the system reads the content of a specific Scrapbox project and generates a response based on that content. If the prompt is a question, it generates an answer. If the prompt is "summarize for me," it generates a summary. If the prompt is "write about XX," it generates content based on that tkgshn.icon. It's like magic!