Surprisingly good litmus test for politick is whether a {pre/post}training researcher immediately blames {post/pre}training for issues rather than defaulting to shared or self blame.
OpenAI research VP Aidan Clark says blaming opposing training phases for technical issues indicates team politics
Mark Chen joked that midtrainers always get the blame.
Users ridiculed the AI researcher's litmus test for pre-post training team politics as amateurish and out of touch, urging him to fix his own issues instead.
No Digg Deeper questions have been answered for this story yet.
Most Activity
@_aidan_clark_ It’s always the midtrainer’s fault
Surprisingly good litmus test for politick is whether a {pre/post}training researcher immediately blames {post/pre}training for issues rather than defaulting to shared or self blame.

@immortaldip Everyone who actually works at frontier labs understands that data doesn’t matter.

@_aidan_clark_ nobody blamed on data team, LOL ?

@_aidan_clark_ So where do distillation attacks fit if “data doesn’t matter”?
Or are traces fundamentally different from data when they’re used for distillation?
@_aidan_clark_ clearly post training was to blame is the right answer
Surprisingly good litmus test for politick is whether a {pre/post}training researcher immediately blames {post/pre}training for issues rather than defaulting to shared or self blame.

@_aidan_clark_ @immortaldip ??💀

@_aidan_clark_ Amateur. Obviously the right move is always blame the infra.

@_aidan_clark_ You mind popping out of your bubble, and applying that puffed out chest to fixing your shit:
@_aidan_clark_ what do the midtraining people do
Surprisingly good litmus test for politick is whether a {pre/post}training researcher immediately blames {post/pre}training for issues rather than defaulting to shared or self blame.