I am looking for service which should exist in AI field. Cloud provider has built the training platform, notebook as service, model as service etc. What other service should exist in this landscape.
Ask HN: Seeking Ideas for next AI service
1–5 of 5 posts
Re: Ask HN: Seeking Ideas for next AI service
#2A cloud platform which allow to get access to very high end training (e.g 20 connected TPUs, A100, etc) with a pay for what you use.
As an AGI researcher I have a few experiments to do on such hardware but just for one to a few days. I would do it if the cost was low.
Re: Ask HN: Seeking Ideas for next AI service
#3A cloud platform which allow to get access to very high end training (e.g 20 connected TPUs, A100, etc) with a pay for what you use. As an AGI researcher I have a few experiments to do on such hardware but just for one to a few days. I would do it if the cost was low.
What's lacking in Google Collab, AWS, GCP and Azure? In my non-ML programmer point of view what you're saying is already available. Why do you think it's not?
Isn't it true that the most high end hardware is available only in these clouds anyways?
Re: Ask HN: Seeking Ideas for next AI service
#4Generally, I advise people not to look for "AI" ideas, look for problems that people have that might be improved or solved using machine learning. Unless you are specifically talking about problems that ML researchers and engineers have, in which case, you should work as one and build solutions like they would today, and see what problems you run into that can be improved.
Re: Ask HN: Seeking Ideas for next AI service
#5A cloud platform which allow to get access to very high end training (e.g 20 connected TPUs, A100, etc) with a pay for what you use. As an AGI researcher I have a few experiments to do on such hardware but just for one to a few days. I would do it if the cost was low.
What's lacking in Google Collab, AWS, GCP and Azure? In my non-ML programmer point of view what you're saying is already available. Why do you think it's not? Isn't it true that the most high end hardware is available only in these clouds anyways?
+1. I'm not sure exactly what the issue is here being described, but pay-per-use, running of large scale ML training techniques across high end instances can be done today. Maybe it's not cheap enough for this person, but if that's the case, I don't really expect a startup to find ways to make it cheaper (unless they subsidize with VC money, but that's not sustainable).