@mupophi@netcapgirl How did that work out for Google and Tesla, who are both building H100 datacenters as fast as possible (despite making TPUs and Dojo)?
Turns out building GPUs and great software to run them hand-in-hand is much more difficult than most people think.
@ocolegro Of course! I've got some compute I can throw at this as well. I've been curious how philosophy/logic/physics reasoning would impact coding focused models. Might need to bifurcate the example styles between coding/non-coding for that.
@audrow Would love to see more easy to use simulation environments that kids can play with. Something like NVIDIA's Isaac Sim but for the roblox generation. Experiment in sim, then a kit for real.
@ylecun easy to talk down about others not open sourcing when you yourself are banking millions/yr from one of those dystopian Silicon Valley companies.
@151eth @jamesbridle@guardian There have been no major breakthroughs in the academic discipline of physics for a couple of centuries. The underlying instrument of math - mainly calculus and linear algebra - were theorized and put into practice back in the 1670s.
@andrewwhite01 Judging by the compensation of top execs of this supposed non-profit, I'm gonna say some cold hard cash... And apparently a social club membership or four.