11–20 of 174 posts
Earlier quoted context omitted.
yes. and the next question is 'ok, how do we add'
I guess that if the bulk of the computation goes into the multiplications, you can work in the log-space and simply sum, and when the time comes to actually do a sum on the original space you can go back and sum.
Isn't this just taking advantage of "log(x) + log(y) = log(xy)"? The IEEE754 floating-point representation stores floats as sign, mantissa, and exponent -- ignore the first two (you quantitized anyway, right?), and the exponent is just an integer storing log() of the float.
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.
Extraordinary claims require extraordinary evidence. Maybe it's possible, but consider that some really smart people, in many different groups, have been working diligently in this space for quite a while; so claims of 95% savings on energy costs _with equivalent performance_ is in the extraordinary category. Of course, we'll see when the tide goes out.
For he sake of the climate and environment it would be nice to be true.
Bad news for Nvidia. “Sell your stock” bad.
Does it come with a demonstration?
Extraordinary claims require extraordinary evidence. Maybe it's possible, but consider that some really smart people, in many different groups, have been working diligently in this space for quite a while; so claims of 95% savings on energy costs _with equivalent performance_ is in the extraordinary category. Of course, we'll see when the tide goes out.
If I could have a SWAG at it I would say a low resolution model like llama-2 would probably be just fine (llama-2 quantizes without too much headache) but a higher resolution model like llama-3 probably not so much, not without massive retraining anyways.
https://arxiv.org/abs/2410.00907 ABSTRACT Large neural networks spend most computation on floating point tensor multiplications. In this work, we find that a floating point multiplier can be approximated by one integer adder with high precision. We propose the linear-complexity multiplication (L-Mul) algorithm that approximates floating point number multiplication with integer addition operations. The new algorithm c…
Presumably there will be a lot of interest.
I posted this about a week ago: https://news.ycombinator.com/item?id=41816598 This has been done for decades in digital circuits, FPGA’s, Digital Signal Processing, etc. Floating point is both resource and power intensive and using FP without the use of dedicated FP processing hardware is something that has been avoided and done without for decades unless absolutely necessary.
Extraordinary claims require extraordinary evidence. Maybe it's possible, but consider that some really smart people, in many different groups, have been working diligently in this space for quite a while; so claims of 95% savings on energy costs _with equivalent performance_ is in the extraordinary category. Of course, we'll see when the tide goes out.
https://arxiv.org/abs/2410.00907 ABSTRACT Large neural networks spend most computation on floating point tensor multiplications. In this work, we find that a floating point multiplier can be approximated by one integer adder with high precision. We propose the linear-complexity multiplication (L-Mul) algorithm that approximates floating point number multiplication with integer addition operations. The new algorithm c…
Does this mean you can train efficiently without GPUs? Presumably there will be a lot of interest.
You still need the GPU parallelism though.