News for hackers. Stuff that matters.
Re:Scalable MatMul-Free Language Modeling by cpldcpu on Sunday June 09, 2024 @12:09PM (#40625409)
>This reaches demoscene levels of crazy/impressive!

The exp/log trick to multiply with addition does indeed look very familiar. I know that a number of demos used it in the 90ies.

Bill Dally, the Nvidia Chief Scientist also seems to be a big fan of log8 representations: https://youtu.be/gofI47kfD28?t=1953

It seems it did not make it into the hardware though, yet...

There are no comments on this comment yet.