pretty sure 10 trillion parameters is now the norm among closed ai labs, given t... | Hacker News

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

alightsoul 21 days ago | parent | context | favorite | on: Cerebras CS-4

pretty sure 10 trillion parameters is now the norm among closed ai labs, given that nvidia also references the same 10 trillion number for their nvl72 racks

KronisLV 21 days ago [–]

Pretty bad efficiency then unless that only applies to Fable class but even then - Kimi K3 is around 3T and does similarly well in most benchmarks.

esafak 21 days ago | [–]

As others have noted, most benchmarks stress the torso, not the tail.

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