buildbot

32 ms·

661.
by buildbot 2y ago
These exist, I have one, and it lets you do exactly what this paper does! https://thinklucid.com/product/triton-5-mp-polarization-came... https://thinklucid.com/polarized-camera-resource-center/
662.
by buildbot 2y ago
Oh that’s really great to know about, thank you!
663.
by buildbot 2y ago
I was about to type something similar to this as well! I use passkeys pretty heavily, with iCloud sync. Never had an issue. The only similar issue I can think of is sometimes my Macbook will loose the contents of the on device wallet, inclu
664.
by buildbot 2y ago
Huh, They used the pile - that's a pretty interesting choice for a corporate research team?
665.
by buildbot 2y ago
Huh, They used the pile - that's a pretty interesting choice for a corporate research team?
666.
by buildbot 3y ago
Does this support training on Apple silicon? It’s not very clear unless I missed something in the README.
667.
by buildbot 3y ago
Yep, this would impact basically any dev at Microsoft, Apple, Facebook, etc. From my own point of view (at MS), policy making decision do basically happen at my/our level too, unless policy is on the tier of "buyout Infection.AI t
668.
by buildbot 3y ago
Brainwave is/was very successful? Still serving your searches! (I work on one of the teams that came out of brainwave, not with them directly) https://www.microsoft.com/en-us/research/project/project-bra
669.
by buildbot 3y ago
Interestingly, reddit seems have gone down about 30-40 minutes ago too.
670.
by buildbot 3y ago
It might be large enough Q2 won't impact performance too much (not guaranteed), but that's a great point. Time to invest in a Milan-X! (Already have the 512GB of DDR4)
671.
by buildbot 3y ago
Quick thoughts - Major arch changes are not that major, mostly GQA and tokenizer improvements. Tokenizer improvement is a under-explored domain IMO. 15T tokens is a ton! 400B model performance looks great, can’t wait for that to be released
672.
by buildbot 3y ago
In most cases they are not, but you need to have a good relationship with your parents and they need to care enough to bother the school about it. Otherwise, they won’t listen to you, a child.
673.
by buildbot 3y ago
I don’t understand how a group allegedly tied to Microsoft AI according to their paper is able to have their own twitter and do their own PR?
674.
by buildbot 3y ago
If you read the article, the central claim is that they failed at making a more efficient model, not a larger, better, GPT-5 model.
675.
by buildbot 3y ago
Unfortunately this specific blog spammer has convinced many a person they are the One True God of the correct path to AGI.
676.
by buildbot 3y ago
Then why is cheaper per token, and following the same path as GPT3 with several cheaper turbo versions.
677.
by buildbot 3y ago
Your understanding is correct, Gary Marcus is intentionally misrepresenting what GPT-4 turbo is to make a story. It’s clearly the cost optimized version of gpt-4. That’s why it’s much cheaper per token.
678.
by buildbot 3y ago
That's not IP, that's secrets. IP needs to be included.
679.
by buildbot 3y ago
> We also review the source for IP/security concerns before releasing. Any modifications you make need to also be open sourced.
680.
by buildbot 3y ago
Good! Companies need to be called out for this more. For example, Phase One, a very high end digital camera company, started using Linux in their digital backs. They were completely unaware of the impacts of GPL!!! At first they refused, un
681.
by buildbot 3y ago
I went to a conference focused on self driving cars back in 2019 - even back then, the best talk was from some professor in Florid, who's main point was that you probably don't have an alligator crossing the road in your training
682.
by buildbot 3y ago
I suppose it might, there are not a lot of details (what kind of sparsity for example?) about what they mean in terms of INT8 support - it could be MXINT8, or something else. Glad someone was thinking the same thing I was though!
683.
by buildbot 3y ago
Hmm interesting, I was linked to an OCP dropbox with a version that did have the connector pinouts. Maybe something someone shouldn’t have posted then…
684.
by buildbot 3y ago
The SXM2 interface is actually publicly documented! There is an open compute spec for a 8-way baseboard. You can find the pinouts there.
685.
by buildbot 3y ago
Probably not, I have had the same experience with a 780m and a Mi60...
686.
by buildbot 3y ago
Fair point!
687.
by buildbot 3y ago
It will be fiber, Ethernet is just the protocol not the physical interface.
688.
by buildbot 3y ago
200gb is not going to be using CAT, it will be fiber (or direct attached copper cable as noted by dogma1138) with a QSFP interface
689.
Making AI More Accurate: Microscaling on Nvidia Blackwell(youtube.com)
2 points by buildbot 3y ago|0 comments
690.
by buildbot 3y ago
Ooff this is a lot of work vs. using an external firewire to sata enclosure - that's what I went with for my powermac G4. But very cool!

More ›