As the title suggests: I'm trying to build a chatbot which his goal should be sort of like "chatgpt".

The chatbot will be installed on Slack workspace and I'm struggling with which scope I should create the documents on for building the index.

At the moment the document scope is "per channel" (i.e. every channel is a whole document), but I'm not sure it's the right approach (maybe it should be far smaller - does it make any difference at all?)

I'm also using huggingface transformers library, but since I'm newbie to this whole new evolving technology, I'm not sure what type of model should I use:

  1. text2text
  2. text-generation
  3. summarization
  4. quiestion-answering

I want the bot to address all (maybe I'm a bit naive).

... so overall, 2 questions:

  1. What should the Document scope be?
  2. Which type of model should I look at? Any specific recommended one?
  • $\begingroup$ Can you explain the usecase of the chatbot you are building so I could help you? $\endgroup$
    – Chay
    Commented Aug 10, 2023 at 23:17


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.