The incredibly clever trick that all frontier AI labs use which makes LLMs so fabulously powerful: they barely regularize at all, resulting in models which are massively overfit. This would be a…

Posters: eshear. eshear

5 likes recorded by the source corpus.