It's actually not the largest. https://huggingface.co/google/switch-c-2048 is 1.... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		zone411 on March 17, 2024 \| parent \| context \| favorite \| on: Grok It's actually not the largest. https://huggingface.co/google/switch-c-2048 is 1.6T parameters.

WeMoveOn on March 20, 2024 [–]

but is switch c even usable? iirc the training set was nowhere near enough for a model of that size to be coherent in a conversation

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact