llm_nerd

95 ms·

1561.
by llm_nerd 3y ago
>The open source community is doing more with GenAI and they're doing it better. Stable Diffusion is extremely cool and is a great project, but it does not compare favourably to Midjourney[1] or Adobe's offering. On the LLM sid
1562.
by llm_nerd 3y ago
This device has a fully switched fabric allowing comms between any of the 256 "superchip" clusters at 900GB/s. That is dramatically faster than a direct host to GPU 32-lane PCI-E connection (which is crazy), and obviously dwa
1563.
by llm_nerd 3y ago
The picture almost looks like a 2U server, but it's actually 24 full server racks . https://www.anandtech.com/show/18877/nvidia-grace-hopper-has...
1564.
Floating Point Numbers used in AI(dennisforbes.ca)
3 points by llm_nerd 3y ago|0 comments
1565.
Understanding Floating Point Numbers(dennisforbes.ca)
2 points by llm_nerd 3y ago|0 comments
1566.
by llm_nerd 3y ago
"The parent comment is trying to dismiss an entire field and also preemptively dismiss new studies" If they're observational or survey based studies, they should be dismissed generally, or at least given incredibly little a
1567.
by llm_nerd 3y ago
That Fortune article was written by the authors of the study itself that they're quoting. The "study" itself was that some people find commutes meditative. I personally do. I'd rather work from home, but it's possib
1568.
by llm_nerd 3y ago
>Are you suggesting that they are not to blame? When they're literally quoting almost verbatim the claims of the researchers, no they aren't to blame. And the vast majority of the time you see an article with some breathless
1569.
by llm_nerd 3y ago
There is a growing trend that when research doesn't hold up (especially in regards to its significance/magnitude/importance), we resort to blaming the journalists who printed the article hyping it. That is seen on here consta
1570.
by llm_nerd 3y ago
This entry very casually goes from describing FP8 and then E4M3 and E5M2, to suddenly talking about FP4. It describes the demonstration of having the mantissa bits "1101" but that by itself is four bits. Unless it's including
1571.
by llm_nerd 3y ago
Discussing a topic rationally and openly isn't a bias, and I've seen no evidence that HN has an anti-car bias particularly. We dedicate a huge amount of the public space to personal vehicles, and it is the sort of solution that ba
1572.
High performance Wikipedia dump parsing with Python(dennisforbes.ca)
4 points by llm_nerd 3y ago|0 comments