Comment Comment by Iv This is actually from a paper published last year:https://www.reddit.com/r/MachineLearning/comments/pffoo8/r_m...A few questions:- Do some ML frameworks implement it already? - It promises up to 200x compression, is it reasonable to expect it to allow us to run GPT-3 on smaller mainstream GPUs? parent Iv 4 years ago
https://www.reddit.com/r/MachineLearning/comments/pffoo8/r_m...
A few questions:
- Do some ML frameworks implement it already? - It promises up to 200x compression, is it reasonable to expect it to allow us to run GPT-3 on smaller mainstream GPUs?