Skip to content

Comment on Groqchatparent

Comments

nomel gave a good answer in a different thread

This is not about the model, it’s about the relative speed improvement from the hardware, with this model as a demo.

To compare apples to apples look at the tokens per second of other systems running Llama 2 70B 4096. We're by far the fastest!

https://news.ycombinator.com/item?id=38742466

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.