-
Notifications
You must be signed in to change notification settings - Fork 55
Description
General Feedback
I built a 120GB VRAM AI rig in my home office. I use AI to help me write professional content. However, the open and closed-source models I have used are ok but not great.
I was an English major in college and have zero math/statistics background.
Just wondering how much of a prerequisite those skills are to be able to learn how to fine tune with h20.ai—and if i invest the time to learn how to use it to fine-tune—will I eventually run into a wall where my lack of math/statistics understanding will cause me to come to a complete stop.
And it goes without saying—I can't just take a night course at a community college. They would force me to start off with algebra and it would be years before I would get to a point where I would have that kind of skill knowledge. Assuming my highly creative brain could even do math at those levels.
As an aside, I really enjoy logic and logic puzzles. I'm hoping that will count for something.
Bonus question: Is 120GB VRAM enough to fine-tune an 8B model? Come hell or high water, I plan on using h20 to at least attempt to fine tune a model—by using either ChatGPT or Claude3 to help me get a basic understanding of what's going on and determine the best settings and approach to the writing challenge(s) that I wish to solve.