tome

35 ms·

571.
by tome 3y ago
Can it be nested? Haskell's do can be used like do r <- do foo bar quux r But I don't see (from https://gleam.run/news/v0.25-introducing-use-expressions/ ) any obv
572.
by tome 3y ago
Point-free means that you don't bind variables, so this particular example is indeed point-free.
573.
by tome 3y ago
"Enshoddification"? (From https://news.ycombinator.com/item?id=39489675 )
574.
by tome 3y ago
Why didn't he email the address provided in the SMS, which will obviously go nowhere else other than to FedEx?
575.
by tome 3y ago
> How would you write what I wrote in Lisp in Haskell main = do let x = 1 let y = 2 let z = 3 let y' = x z' = y x' = z print [x', y', z']
576.
by tome 3y ago
I don't know what that is. It's nothing to do with Groq Inc.
577.
by tome 3y ago
Groq does not run a blockchain and our chips were never intended for crypto mining.
578.
by tome 3y ago
The chips are built for general purpose low latency, high throughput numerical compute.
579.
by tome 3y ago
Yes, sorry about that, it's because of the huge uptick in demand we've had since we went viral. We're building out more and more hardware to cope with demand. I don't think we have any quality of service guarantees for
580.
by tome 3y ago
OK, thanks, that's useful to know. Personally I'm not involved directly in implementing the model, so I don't know what we do there.
581.
by tome 3y ago
Yes, in the future we'd like to do that.
582.
by tome 3y ago
Interesting question! Hopefully being faster is so much more useful to you that you use a lot more :)
583.
by tome 3y ago
We are absolutely slammed with requests right now, so I don't know, sorry.
584.
by tome 3y ago
Here's a live demo of CNN of Groq plugged into a voice API https://www.youtube.com/watch?v=pRUddK6sxDg&t=235s
585.
by tome 3y ago
Yeah, it's confusing. See here for an explanation: https://news.ycombinator.com/item?id=39431921
586.
by tome 3y ago
Currently graphics processors work well for training. Language processors (LPUs) excel at inference.
587.
by tome 3y ago
There is an API and there is a waitlist. Sign up at http://wow.groq.com/
588.
by tome 3y ago
Our interconnect between chips is also deterministic! You can read more about our interconnect, synchronisation, and error correction in our paper. https://wow.groq.com/wp-content/uploads/2023/05/GroqISC
589.
by tome 3y ago
I'm sure you'll hear all about our IPO on HN :) :)
590.
by tome 3y ago
I think the models change far too quickly for that to be viable. A chip has to last several years. Currently we're seeing groundbreaking models released every few months.
591.
by tome 3y ago
I guess that's total throughput, rather than per user? You can increase total throughput by scaling horizontally. You can't increase throughput per user that way.
592.
by tome 3y ago
Yeah it's a bit confusing. See here for details: https://news.ycombinator.com/item?id=39431921
593.
by tome 3y ago
At the moment we're concentrating on building out our API and serving the enterprise market.
594.
by tome 3y ago
Great idea, but Groq doesn't have a product suitable for training at the moment. Our LPUs shine in inference.
595.
by tome 3y ago
Yes, we don't have any publicly accessible ones at the moment though.
596.
by tome 3y ago
Exciting! Looking forward to seeing it.
597.
by tome 3y ago
What quantified testing would you like to see? We've had a lot of very good feedback from our users, particularly about Mixtral.
598.
by tome 3y ago
Yup, understood. Access to consumer hardware like this is not something that we provide at the moment, I'm afraid.
599.
by tome 3y ago
There are a lot of people interested in Groq now, so most jobs are sitting in a queue for a little while.
600.
by tome 3y ago
Yes, we're working with some customers on that, but it will be a while until general availability.

More ›