(For a model of GPT-4's size, it could also be 8 nodes with several GPUs each, e... | Hacker News

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

cjbprime on April 19, 2024 | parent | context | favorite | on: Meta Llama 3

(For a model of GPT-4's size, it could also be 8 nodes with several GPUs each, each node comprising a single expert.)

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