News for hackers. Stuff that matters.

A comment on Llama 3.1 Official Launch.

Re:Llama 3.1 Official Launch by IceHegel on Tuesday July 23, 2024 @03:27PM (#41049811)
Will 405b run on 8x H100s? Will it need to be quantized?
[ Parent | on HN ]
Re:Llama 3.1 Official Launch by bddppq on Tuesday July 23, 2024 @08:48PM (#41052589)
yep with <= 8bit (int8/fp8) quantization
[ Parent | on HN ]