matanyal

5 ms·

1.
by matanyal 3y ago
Hey y'all, we have a discord now for more discussion and announcements: https://discord.com/invite/TQcy5EBdCP
2.
by matanyal 3y ago
When you have a pile of hardware and silicon Intellectual Property, patents, etc, IMO it's pretty clever. However, I'm a Groq Engineer, and I'm mega-biased.
3.
by matanyal 3y ago
We intend to be very competitive on cost, power, hardware, TCO, whatever it is. Custom-built silicon+hardware has the advantage in this space.
4.
by matanyal 3y ago
Thanks! Feel free to join our discord for more announcements and demos! https://discord.com/invite/TQcy5EBdCP
5.
by matanyal 3y ago
We are great for image embedding (and audio, with more to come!) There is no reason you should be forced to use graphics cards intended for gaming for any AI workload.
6.
by matanyal 3y ago
Thanks!
7.
by matanyal 3y ago
Groq Engineer here as well; we actually built our compiler to compile pytorch, TensorFlow, and Onnx natively, so a lot of the amazing work being done by y'all isn't building much of a moat. We got LLama2 working on our hardware in
8.
by matanyal 3y ago
https://groq.com/groq-sets-new-large-language-model-performa...
9.
by matanyal 3y ago
More info about Groq: https://groq.com/lpu-inference-engine/
10.
by matanyal 3y ago
Reading is one thing, but think about stuff like website generation, searching for information in massive datasets, real-time audio chats that don't sound like the AI misheard everything with a pause, and stuff like that.
11.
by matanyal 3y ago
It told me "Yeehaw Ridem Cowboy" was potentially problematic, which is news to me living out west.
12.
by matanyal 3y ago
Interesting, no mention of Groq for number 6.
13.
by matanyal 3y ago
Hey Jeff, Good to see that y'all are still pushing in the right direction! I left between the layoffs to pursue some innovation in the semiconductor space, but I look back fondly on my time at Twilio (Sendgrid)! Building APIs into busi