Or are the current algorithms just failing on darker faces because of some inherent issue? shadows/lighting?edge detection?
It doesn't seem that hard to make a diverse dataset, so why are most of the models on the market awful at it?
GPT 3.5 turbo can do it pretty well, but it's out of my budget, as I would need to make a LOT of requests.
I have about 8,000$ in AWS credits for self hosting
Is there any LLM you'd recommend for this?
I was thinking of reverse engineering copilot and using its internal API directly, do you know if there is a big rate limit?
thanks
But I just had an idea that seems really great to me, and potentially really profitable, really quickly, if I move my ass to launch an MVP. There is a market for it that's 100% sure, if my product is good enough and well marketed it can definitely succeed.
I was about to ask for feedback about it to an alumnus from my school, who's an industry expert.
But I took a step back and I was like: he was in school a year before me, he have more business experience, he's a better seller, have a bit more technical experience overall, launched a few successful projects, more domain expertise, a better network, more capital.
If I expose my project to him, and he's actually interested, he actually have a lot more chance of succeeding than me, and he definitely don't need me, so I would have just lost my first mover advantage.
My only real advantage would be one specific technical skill that he don't have, but he very likely have the means to outsource this part.
Do you think you should never be afraid to share your business ideas? Or are there specific cases where it's just a terrible idea to do so?
