chaxor

35 ms·

361.
by chaxor 3y ago
Is not even that - this entire idea just misses the point entirely. It doesn't matter if a very particular model $A$ under particular conditions can be identified with another particular model $C$. That can very easily be circumvent
362.
by chaxor 3y ago
This is probably a stupid question regarding the equipment, but it seems to be a good addition for the user, even if additional hardware underneath is required - why is CO not more common on these devices? Additionally, where are all the pp
363.
by chaxor 3y ago
Cool. So when do we get whatever science psychology will turn into?
364.
by chaxor 3y ago
Ah, so completely unusable. Got it.
365.
by chaxor 3y ago
It can do well with optimization and readability *if you ask it specifically for those things*. Especially if you have a particular paradigm and algorithm in mind (you obviously already should anyway). This is why these systems are helpful
366.
by chaxor 3y ago
Deepmind has done quite an enormous amount actually, but it's been in academia not in the commercial product sphere. Just because something is not on a little web page available to average Joe's does not mean there isn't va
367.
by chaxor 3y ago
This is unbelievably wrong. Deepmind has probably the best academic group in RL. The difference between Deepmind and OpenAI is that Deepmind favors academic endeavours and novelty much more, while completely ignoring any commercialization
368.
by chaxor 3y ago
This problem is more far reaching than just extension, and further reaching than what entity is in charge of something. For instance, the worst company imaginable may be in charge of software that was once FOSS, and they may change absolute
369.
by chaxor 3y ago
To quote Lukaz Kaiser in his famous talk on 2017: "stupidity is all you need"
370.
by chaxor 3y ago
Why is commercial necessary to run local models?
371.
by chaxor 3y ago
This has been done for a very long time. Blockchains are definitely not required (this isn't just the usual hate from HN of Blockchain, it just actually doesn't make sense here). Fingerprinting by shingling (windows of text) wit
372.
by chaxor 3y ago
I'm not so sure that there's enough evidence in any cases that there is an improvement on health or outcome. Some domains, absolutely. However, many areas we have made very little progress in either mechanistic understanding, or t
373.
by chaxor 3y ago
The only real way to block specifically openAI generated content (or some other online generated LLM content) is for the company itself to store all of its outputs and compare to that database, like shingling / LSH for plagiarism det
374.
by chaxor 3y ago
The trick is to hide all extra stuff like address bar, tabs, settings, OS related bars, etc. and alonly show the browser stuff upon Ctrl+L. It's so much nicer, especially with an wm that's just gets out of the way - awesomewm
375.
by chaxor 3y ago
Hide all of it - all things not the page itself. The only time the address bar, settings accessibility, etc should be visible is upon Ctrl+L
376.
by chaxor 3y ago
That's fair, an encoder system for which the full attention values and weights available is a useful tool. IMO encoder models are far more powerful and useful for devs than any of the decoder models.
377.
by chaxor 3y ago
I have gotten that quite a few times now and it does seem to be new. In my case though, it acted as if it would change something, but then didn't, and got stuck in a loop. I got it enough times that it seemed to happen more often when
378.
by chaxor 3y ago
This language bothers me deeply. So many people ask the system itself to describe what it is doing, for which it is absolutely incapable of doing. The system has no access to its internals. Even academics are asking the LLM itself to descri
379.
by chaxor 3y ago
I think they're trying to say there is a good use case for everything. For instance, if you have a set of keyword-like IDs (normalized items like EntityID:47283 for example) for a container of these things, like documents, you can make
380.
by chaxor 3y ago
This is ridiculous, obviously flawed in its efficacy. My way is way better - just simply look at the entire code base, read every file, make sure nothing phones home, internally criticize the design choices, realize I could have written som
381.
by chaxor 3y ago
That's very nice and I would like to see it really take off. Do they use the new toroidal propellers for better efficiency and lower noise?
382.
by chaxor 3y ago
The reason it's worse is basically because it's more 'safe' (not racist, etc). That of course sounds insane, and doesn't mean that safety shouldn't be strived for, etc - but there's an explanation as to h
383.
by chaxor 3y ago
Google (Deepmind) actually has the people and has developed the science to make the best AI products in the world, but unfortunately Bard seems to be thrown together in an afternoon by an intern, and then handed off to a hoard of marketing
384.
by chaxor 3y ago
Sure ok, there's a bit of namespace conflict now, happens quite a lot. I also don't love it, but it's there now, and it's really not the worst name. It's a good area though, and the paper seems interesting. Do yo
385.
by chaxor 3y ago
The term classifies a different type of characteristic of the learning trajectory, so just using 'learning' would be missing that distinction. There needs to be a separate term for 'long stretch of epochs for which there is a
386.
by chaxor 3y ago
What is untrusted code? Everyone has different threat models. Some may treat Linux or BSD as untrusted.
387.
by chaxor 3y ago
This is a misguided conclusion, as they are entirely different systems. Don't compare apples and oranges like this or you'll be very disappointed.
388.
by chaxor 3y ago
I tend to agree, though with perhaps a much larger tolerance in the number of years "not far off" is. What may be interesting is seeing the emergence of hardware specific bootloader's, kernels and OS, rather than using boot&#
389.
by chaxor 3y ago
A 'Tree of thoughts' if you will...
390.
by chaxor 3y ago
Features are the problem. Stop adding features. Make packages as best you can, such that, If you really want something added on, it's a separate package to add on.

More ›