yyyk

35 ms·

481.
by yyyk 3y ago
>One's person expensive computation is another person's almost free computation We have pretty good guides to what's an expensive computation to everyone. That's how password-hashing algorithms work, much better than
482.
by yyyk 3y ago
Archive.org is a bit of a special case, you need to call them repeatedly to archive a website. They do have a rate limit there, it's pretty aggressive* to the point you could trip it by manually using the site. They must have forgotten
483.
by yyyk 3y ago
Self driving cars can be effectively perfect - if we finally stop designing cars that cater to human drivers and stop trying to look like current cars. IMHO, if we finally accept that AI cars are different, we can achieve that level today
484.
by yyyk 3y ago
I couldn't make the comparison since I upgraded my internet connection at the same time. I strongly suspect most (all?) of the improvement came to that.
485.
by yyyk 3y ago
IIRC, There are studies which show a difference, especially in poor latency scenarios. The issue with HTTP/2 is that it complicates the stack: ideally one would support either HTTP/2 or HTTP/3 but there's little point in
486.
by yyyk 3y ago
Thinking about this a bit more, the main issue is that the LLM is an attack vector. Hallucinations aren't a big deal if the action requires approval or can be easily undone. The only way right now I can see to make LLM interfaces &#x27
487.
by yyyk 3y ago
The issue is economic policy, mainly Erdogan's distaste for high interest rates and results thereof. Mind you, Egypt has different serious economic problems, but these are more long term issues...
488.
by yyyk 3y ago
"Maybe READ COMMITTED is good enough or maybe people don’t know how dirty their data actually is…" Look, a deadlock or slowness may technically be correct, but it can mean a lost sale. In practice, a worse problem than data incons
489.
by yyyk 3y ago
>If only there were a way to mark a section of HTML as "no scripting here" iframe sandbox with srcdoc? Not so elegant, but it works. [Alternatively, iframe to another document and use CSP header there to ban everything]
490.
by yyyk 3y ago
>was specific to Acinetobacter baumannii We can reasonably expect the bacteria to mutate against the new antibiotic if/once it's used. It's one shifty opponent. This may make the model obsolete, but maybe not - there'
491.
by yyyk 3y ago
> the only way to get apps from the Google Play Store There's also Raccon: https://raccoon.onyxbits.de/
492.
by yyyk 3y ago
The first sentence I quoted. They used thousands to train the AI, and then ran the AI on a similar sized list. Perhaps with reuse it will be worthwhile. Or the second list was harder to synthesize?
493.
by yyyk 3y ago
>They took thousands of drugs where the precise chemical structure was known, and manually tested them on Acinetobacter baumanni... This information was fed into the AI... The AI was then unleashed on a list of 6,680 compounds... took th
494.
by yyyk 3y ago
Well, I guess we agree here - the basic functionality will be common, Open AI will not have a monopoly (unless lawmakers really screw up), but will likely exist and be a major player. I am not sure about Open Source at all just yet - e.g. i
495.
by yyyk 3y ago
>True, but every system has flaws. For the most part the Euro and ECB has been a benefit for every country in it. Except Italy alas.
496.
by yyyk 3y ago
I agree OpenAI won't and can't have anywhere near a monopoly, excluding very unlikely legal shenanigans. The field is too approachable, and ultimately, to pinch into their market one doesn't need GPT-4 level performance. Besi
497.
by yyyk 3y ago
The logo is nice. The bigger news is the expected iOS thunderbird version. Would it be new code or based on some existing iOS mail program (like the Android version)?
498.
by yyyk 3y ago
For example, the error could have been '1064 is not a string value' or at least 'float64 is not a string value' etc.
499.
by yyyk 3y ago
What you refer as 'the underlying code' isn't the real underlying code, the underlying code is the model. Your underlying code is just the runtime, which is commercially the least interesting part (which is why it is/wil
500.
by yyyk 3y ago
The ability to talk about 'forbidden' topics is also a significant dis advantage. Just wait for the first moral panic targeting open source GPTs. I think that Open Source will exist (barring a legal ban), but the triumphalism is
501.
by yyyk 3y ago
My point is that you know the real source of the error, so there's no point for these terse messages. The example in the article, you know at point of conversion that you get a float, but instead you need to do a dance to get the origi
502.
by yyyk 3y ago
>Open models... are still behind GPT-4 by some distance but they are closing in at a rate that leads me to believe there’s not much of a moat there Commercially, this doesn't matter even if true. As I keep reminding here: moats are
503.
by yyyk 3y ago
I won't take him seriously on this - if one really believes in the case for regulation, than there's no good reason to exclude open source models. Let's take the most pessimistic possibility, that they have a slower rate of p
504.
by yyyk 3y ago
"Wouldn’t it be great if you could see errors in place instead of mysterious NULLs?" It's not a good ad when the error message is inadequate even in the supplied example and you need to hack around it.
505.
by yyyk 3y ago
This is a significant announcement. Note namechecking every non-Apple CPU vendor and NVIDIA - but not mentioning AMD/Intel GPUs. Another note: Lots of small improvements, but not allowing users to move the Taskbar (to Microsoft, some t
506.
by yyyk 3y ago
>...she never actually tries her recipes. When it's time to do a new piece, she sits down at her computer and makes up a recipe. She is experienced and knowledgeable she it usually turns out ok. If she's experienced I guess tha
507.
by yyyk 3y ago
AFAICT it uses steamship_langchain which calls to OpenAI. It's technically 'OSS' in the sense a client to a closed system is OSS.
508.
by yyyk 3y ago
You've given me some things to think about. I see how your experiment would disprove Poverty of Stimulus (if we could roughly assume how much input a baby gets) but I don't see how it would disprove Universal Grammar. If the compu
509.
by yyyk 3y ago
So if we take the Universal Grammar argument - not merely some specialization but specialization towards underlying structure - I don't see how your experiment would disprove that. Chomsky could always argue the model had 1000x the pro
510.
by yyyk 3y ago
>why do you need a Google Drive when your ChatGPT can also store files for you? If the idea is to keep the actual files rather than summaries, than there's an entire world of requirements (data storage reliability, access control, a

More ›