chaxor

38 ms·

511.
by chaxor 3y ago
Totally agreed. Punch cards are such an arduous technology to use, they're effectively useless. Therefore it's pretty clear to me computation is not really going anywhere as a technology and I highly doubt much more progress can
512.
by chaxor 3y ago
There should be a very simple way to translate Matlab to Julia that anyone can use easily. Yeah, GPT-4 "could* do that - but I don't think very many people would ever do that. Why? Because often Matlab code is private and the own
513.
by chaxor 3y ago
There's a decent working paper that has benchmarks on this, if you're interested. There are many types of reasoning, but GPT-4 gets 97% on casual discovery, and 92% on counterfactuals (only 6% off from human, btw) with 86% on actu
514.
by chaxor 3y ago
It's possible that the comment you're responding to is referring to the new paper out of Stanford/Google on emergence as a 'mirage'. The gist of it is that the shape of the curve for 'metric' vs 'para
515.
by chaxor 3y ago
I hear crushed glass has very great health benefits
516.
by chaxor 3y ago
A decade can still be quite short considering the types of problems that may need to be solved, so perhaps that is the timescale that some of the people you're referring to are thinking in?
517.
by chaxor 3y ago
It seems a positive control is missing in these studies. Perhaps humans (not a perfect solution, but unfortunately the best 'learning system' available) could be used as a control?
518.
by chaxor 3y ago
Having excellent semantic based compression is an enormous technological advancement. The fact that petabytes of textual information - effectively the entire history of the world in written text can be compressed to just a few hundred gig
519.
by chaxor 3y ago
This type of thinking really needs to look beyond what the hype train is on with OpenAI right now and look at some of the more groundbreaking work going on. Deepmind has quite a bit of work outside the language area that has a lot of potent
520.
by chaxor 3y ago
Qdrant is *far* better than pinecone. There's really no comparison.
521.
by chaxor 3y ago
Yeah, that's an option I'm aware of, but it gets quite ugly when the rules are more than just a few.
522.
by chaxor 3y ago
Yeah, those are excellent choices when purchasing, but when you already have something and are trying to use it - you should be able to.
523.
by chaxor 3y ago
Well, this isn't necessarily about hacking - it's about backing up a phone too. Google doesn't allow backing up a phone in any reasonable way.
524.
by chaxor 3y ago
That's pretty wild, I've been setting things up like this for about 5 years with just BERT or my own fine tuned encoder only systems. It should be done for free, not millions... Can I get millions for running `ls` too?
525.
by chaxor 3y ago
One issue for this process that I have seen when talking about the Pixel is: - ***You can't actually back up your Pixel*** It's enormously frustrating, because this is the typical advice before moving to a different OS.
526.
by chaxor 3y ago
It could be from training if more to be safer. This was noted by Microsoft early on with GPT-4. Specifically, when looking at the tikz unicorn qualitative benchmark, the unicorn got better with more epochs, which is obviously expected. Howe
527.
by chaxor 3y ago
This is a decent way to explain a good use case - exploration of new priors and ideas. For example, I was looking for a good way to do streaming xml parsing in the simplest way possible. One way is to do it via various packages in different
528.
by chaxor 3y ago
Everytime I see pinecone I'm immediately put off by it since it's so heavily focused on "use the cloud" and "make an account". I don't want another account, I don't want to have to be connected to the
529.
by chaxor 3y ago
In what world have web devs ever been able to look down on real software engineers?
530.
by chaxor 3y ago
"steers the project in a different direction" Man, it can't read minds still? Literally unusable.
531.
by chaxor 3y ago
I'm actually not so sure about that statement. For example, knowing if the code will be executed on a raspberry pi, a HPC with 10TB RAM and 512 CPUs, or a home desktop with 128GB RAM and 8 core CPU will greatly affect how the task may
532.
by chaxor 3y ago
I love this comment, thank for such a good laugh. I really hope no one would ever think this non sarcastically.
533.
by chaxor 3y ago
Many people are using it, but people that are relying on OpenAI to do all of the work for them aren't looking deep enough to realize it.
534.
by chaxor 3y ago
Wireguard has a pre-shared key that can essentially get 'added to' the base key, making it more secure. Rosenpass effectively just makes these PSKs and trades them in a way that makes it quantum secure. Basically it should be a p
535.
by chaxor 3y ago
I have a team of bioinformaticians with very little ML knowledge, but they know of these basic papers... So yeah, if someone claims to be in ML and aren't aware of this foundational knowledge... they're not to be taken seriously.
536.
by chaxor 3y ago
The reason they don't know them is they're not serious researchers or practitioners of ML - it's as simple as that. Anyone in this area should have this exceedingly basic common knowledge.
537.
by chaxor 3y ago
PhD students are typically more knowledgeable than the PI, so this is probably optimal. Perhaps a good Postdoc is better. PIs are often way outdated in their knowledge and don't have the time or energy to stay abreast the development
538.
by chaxor 3y ago
The joke photo is much closer to reality than many ight believe, according to OpenAI at least. OpenAI employees have claimed they can search direct strings generated by the system for expanding bullet points on their whole database and rece
539.
by chaxor 3y ago
Seems extremely similar to the Point-e work done a few months ago. I would imagine that just doing Veitoris Rips on the point-e would get close to this solution as well, although that would be a bit too logic based for the ML world we live
540.
by chaxor 3y ago
Of the goal is to use it to access a large knowledge base (like Google, but with better semantic searching), then it doesn't matter as much . There are some cases where it still matter due to not making some connections (for example,

More ›