Llama 3.1 Official Launch

Re:Llama 3.1 Official Launch by quotemstr on Tuesday July 23, 2024 @12:18PM (#41047742)

Groq's TSP architecture is one of the weirder and more wonderful ISAs I've seen lately. The choice of SRAM in fascinating. Are you guys planning on publishing anything about how you bridged the gap between your order-hundreds-megabytes SRAM TSP main memory and multi-TB model sizes?

[ Parent | on HN ]

Re:Llama 3.1 Official Launch by foundval on Tuesday July 23, 2024 @12:24PM (#41047848)

There is a lot out here.

I gave a seminar about the overall approach recently, abstract: https://shorturl.at/E7TcA, recording: https://shorturl.at/zBcoL.

This two-part AMA has a lot more detail if you're already familiar with what we do:

https://www.youtube.com/watch?v=UztfweS-7MU

https://www.youtube.com/watch?v=GOGuSJe2C6U

[ Parent | on HN ]

Re:Llama 3.1 Official Launch by quotemstr on Tuesday July 23, 2024 @12:26PM (#41047863)

Thanks!

[ Parent | on HN ]