ta988

39 ms·

751.
by ta988 3y ago
George Hotz (pseudo geohot) in his recent interview with Lex Fridman gave some info on the probable structure of gpt4.
752.
by ta988 3y ago
Especially if you willingfully used data that they shouldn't have shared to you in the first place.
753.
by ta988 3y ago
It would be easier if the tax code was simplified, oh wait the tax-prep companies are also lobbying against that. Maybe we should dismantle them, take their money away and use that to pay lawyers and tax specialists to simplify the tax code
754.
by ta988 3y ago
And everybody working for ad companies is complicit.
755.
by ta988 3y ago
I really enjoy their new UI, it is so much more practical and spacious.
756.
by ta988 3y ago
That's interesting because that's more or less on more level above the multi-head attention.
757.
by ta988 3y ago
So every broke artist that cannot afford the fees would not benefit from their work once it becomes recognized way later in their life? that sound pretty bad.
758.
by ta988 3y ago
ghee is refined butter and you could argue butter is refined cream. Those definitions are all fuzzy and context-dependant.
759.
by ta988 3y ago
I agree, that's how telemetry should be done.
760.
by ta988 3y ago
It was a joke, langchain was released last year.
761.
by ta988 3y ago
So is it about trying to mimic what the free software devs do and hope that it will provide the same outcome? Sounds like dressing like your favorite rock star to try to play guitar better.
762.
by ta988 3y ago
There are specialized reddit forums about that where you can give your budget and preferences and people help you with picking the right parts. They will give you a pcpartpicker list or you can start giving one and they will tell you if it
763.
Fedora 40 Considers Telemetry(phoronix.com)
18 points by ta988 3y ago|6 comments
764.
by ta988 3y ago
This has been going on for months, and the rewrite will just introduce new bugs. That doesn't sound like the correct solution to a data integrity problem.
765.
by ta988 3y ago
Yep had a lot of bad issues because of that too. I'm currently looking for another provider where I can use reliable solutions (just a real email client that does imap/smtp and handles gpg).
766.
by ta988 3y ago
What does it do?
767.
by ta988 3y ago
Yep I have a small shell script that caches a repo with only the last commit files and runs ripgrep on it. I'll give a go at live grep that's discusses above, that look exciting.
768.
by ta988 3y ago
Does it support the video/screen sharing/huddles now?
769.
by ta988 3y ago
You still get a delay when switching desktops.
770.
by ta988 3y ago
They are pivoting toward industrial and higher value applications as well.
771.
by ta988 3y ago
GPT is not understanding language we have many examples that show that. It has some other qualities that make it good at interacting with humans using language, but it is never "understanding". That's one part of the points t
772.
by ta988 3y ago
One of the authors is a philosopher (mostly realist, logic and ontologist) so that's a different angle than a math based book that's for sure. And that's also what makes it interesting compared to most other books that went o
773.
by ta988 3y ago
There has been some work on that recently see: https://www.routledge.com/Why-Machines-Will-Never-Rule-the-W... It is dense and opinionated but even if you may disagree at the end, it is absolutely worth a read.
774.
by ta988 3y ago
You may want to get a chest xray to check for aortic calcification at some point.
775.
by ta988 3y ago
Their robots can't even walk by the sea without being slowly destroyed by salt (and they are not really adapted to sand walking). So we are far.
776.
by ta988 3y ago
It highly depends on the definition you use, gather them from wikipedia and a few dictionaries and you will see. Looking at it from different languages is also worth looking into. I see it as an unavoidable (so far and in my opinion for a l
777.
by ta988 3y ago
Sounds like a rich people trap. You can say anything if you don't show the results. Patents are also not a proof of anything remotely close to working. So for now they have... nothing?
778.
by ta988 3y ago
I found that useful as I had no idea what SDF was: https://voxel-tools.readthedocs.io/en/latest/smooth_terrain/
779.
by ta988 3y ago
There were card sorting machines that were doing exactly that
780.
by ta988 3y ago
As long as you get new data in a way that can falsify your hypothesis that's fine. If you bias your data collection to favor your hypothesis that's still cheating.

More ›