3 ms·
That doesn’t make sense though if scaling is actually stalling. The reason so much compute goes into training now is scaling, which keeps base model lifetime sh
by bjornsing 2y ago
That doesn’t make sense though if scaling is actually stalling. The reason so much compute goes into training now is scaling, which keeps base model lifetime short.