[badFEengineer]

This was surprisingly fast, 276.27 T/s (although Llama 2 70B is noticeably worse than GPT-4 turbo). I'm actually curious if there's good benchmarks for inference tokens per second- I imagine it's a bit different for throughput vs. single inference optimization, but curious if there's an analysis somewhere on this

edit: I re-ran the same prompt on perplexity llama-2-70b and getting 59 tokens per sec there

[andygeorge]

fast but wrong/gibberish

[razorguymania]

Its using vanilla llama-2 from Meta with no fine tuning. The point here is the speed and responsiveness of the underlying HW and SW.

[chihuahua]

But if the quality of the response is poor, it's irrelevant that it was generated quickly. If it was using different data to generate higher quality responses, would that not slow it down?

[tome]

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

[andygeorge]

Do you work there? Just curious

[razorguymania]

yes

[andygeorge]

Thanks!

[andygeorge]

ah Llama 2 70B, no wonder