[StochasticLi]

I would like to know what inference speeds they are achieving exactly on what hardware. I skimmed and searched the article and didn't find that info.