News for hackers. Stuff that matters.

A comment on AI engineers claim new algorithm reduces AI power consumption by 95%. Parent comment

Re:AI engineers claim new algorithm reduces AI power consump... by mota7 on Saturday October 19, 2024 @03:49PM (#41890236)
Not quite: It's taking advantage of (1+a)(1+b) = 1 + a + b + ab. And where a and b are both small-ish, ab is really small and can just be ignored.

So it turns the (1+a)(1+b) into 1+a+b. Which is definitely not the same! But it turns out, machine guessing apparently doesn't care much about the difference.

[ Parent | on HN ]
Re:AI engineers claim new algorithm reduces AI power consump... by amelius on Saturday October 19, 2024 @04:08PM (#41890382)
You might then as well replace the multiplication by the addition in the original network. In that case you're not even approximating anything.

Am I missing something?

[ Parent | on HN ]
Re:AI engineers claim new algorithm reduces AI power consump... by dotnet00 on Sunday October 20, 2024 @01:36AM (#41893129)
They're applying that simplification to the exponent bits of an 8 bit float. The range is so small that the approximation to multiplication is going to be pretty close.
[ Parent | on HN ]
Re:AI engineers claim new algorithm reduces AI power consump... by tommiegannert on Saturday October 19, 2024 @04:25PM (#41890513)
Plus the 2^-l(m) correction term.

Feels like multiplication shouldn't be needed for convergence, just monotonicity? I wonder how well it would perform if the model was actually trained the same way.

[ Parent | on HN ]
Re:AI engineers claim new algorithm reduces AI power consump... by dsv3099i on Saturday October 19, 2024 @09:14PM (#41892121)
This trick is used a ton when doing hand calculation in engineering as well. It can save a lot of work.

You're going to have tolerance on the result anyway, so what's a little more error. :)

[ Parent | on HN ]