Re:Running large language models like ChatGPT on a single GPU
by JoshCole on Monday February 20, 2023 @01:19PM
(#34870955)
No, it isn't astronomical. It is smaller than that. Still large, but not astronomical.
[ Parent
| on HN ]
Re:Running large language models like ChatGPT on a single GPU
by ipsum2 on Monday February 20, 2023 @02:49PM
(#34872067)
Have you tried training a large model before? If not, you're probably discounting how difficult and expensive it is.
Re:Running large language models like ChatGPT on a single GPU
by coldtea on Monday February 20, 2023 @04:56PM
(#34873487)
Well, for those that trained the largest one atm, it cost them in the order of 10 million dollars (actually less).
That's how much some tech companies pay for catering.
Hell, that's in the order of a single socialite's wedding costs.
Re:Running large language models like ChatGPT on a single GPU
by JoshCole on Monday February 20, 2023 @10:19PM
(#34876042)
At Voyager's speed it would take approximately 749,000,000 years to reach Canis Major Dwarf. OpenAI was founded in 2015. So it has been eight years. 749,000,000 - 8 = 748,999,992. 1.06809079e-8% of the time of a random astronomical time; rounded, that is about, uh, 0.00%ish.
I mean, don't get me wrong. It is a very expensive project. It just isn't astronomical. Anyone reading this and thinking - oh I could never do that even in hundreds of millions of years - that would be wrong. If you won the lottery or just made good financial decisions you could do a project comparable to this instead of getting a very nice house in the Bay Area.