DeepLLM, a tool that automates focusing GPT-4's reasoning

First, welcome to the forum, were glad to have you and appreciate your original contribution here.

I read through your paper and while it is certainly intriguing, it falls somewhat short of compelling for me due to the lack of quantifiable metrics.

What I would love to see with this is a proposed benchmark by which you could compare your approach to others—some kind of headline numbers which demonstrate the value of your contribution.

I’m any event, after a quick read I think you’ve got something interesting here.

1 Like