Re:Ethical Challenges Related to the NeurIPS 2024 Best Paper...
by m_ke on Thursday December 12, 2024 @09:31AM
(#42399402)
He's a graduate student with multiple influential papers, including NeurIPS best paper award here that might change how to do image / video generation, I think he can figure out docker and tweak a few lines of python.
Re:Ethical Challenges Related to the NeurIPS 2024 Best Paper...
by brookst on Thursday December 12, 2024 @09:46AM
(#42399540)
It’s more the black hat expertise I’m referring to. Knowing that containers aren’t versioned, and that random seeds are a a good attack vector.
Of course he knows python. But apparently also a very accomplished systems thinker who can apply it to a different domain.
Re:Ethical Challenges Related to the NeurIPS 2024 Best Paper...
by m_ke on Thursday December 12, 2024 @09:48AM
(#42399563)
No it's not, anyone who ever used docker knows that the images don't get committed and setting random seeds is ML 101 for anyone who is trying to run repeatable experiments.
Re:Ethical Challenges Related to the NeurIPS 2024 Best Paper...
by RicoElectrico on Thursday December 12, 2024 @09:50AM
(#42399578)
This all was described before, and acted on by ByteDance.
And let's be real, there's an inflation of what "intern" means in academia, finance and elsewhere like FAANG. They're often doing a job of an FTE in less-prestigious employers, just that it's maybe not directly expected to pay off.
All this as an intern? That’s one multi-talented evil genius if true.