kaoD

19 ms·

331.
by kaoD 2y ago
I can see how you got that impression from my comment (if you ignore how I mentioned architecture), so let me elaborate: It's the opposite. FizzBuzz and getting the syntax right is what LLMs are good at... but there's so much more
332.
by kaoD 2y ago
I'm a senior engineer (as in, really senior, not only years of experience). I can get familiar with unfamiliar APIs in a few hours and then I can be sure I'm doing the right thing, instead of silently failing to meet edge cases an
333.
by kaoD 2y ago
This is how companies fail. 5 years down the line no one is able to change anything in the system because it's so poorly architected (by being a bunch of Claude copypastes cobbled together) that it takes one month to do a one-day task
334.
by kaoD 2y ago
> They should, because LLMs are coming for them also, just maybe 2-3 years later than for programmers that aren't "real top-tier". Would you be willing to set a deadline (not fuzzy dates) when my job is going to be taken b
335.
by kaoD 2y ago
This. Most people I know that use LLMs to be super productive are like "make me a button, it's red" (hyperbolic statement but you know what I mean). I can do that faster and better myself. When I'm deeply stuck on someth
336.
by kaoD 2y ago
> LLMs do now have a concept of truth now since much of the RLHF is focused on making them more accurate and true. Is it? I thought RLHF was mostly focused on making them (1) generate text that looks like a conversation/chat/as
337.
by kaoD 2y ago
Does the commutative property change anything here? A, B and C are not constrained in any way to each other, so they can be in whatever order you want anyways... Moreover, addition is commutative so it doesn't matter what order the div
338.
by kaoD 2y ago
Wait, there are negative rings!? I'd like to learn more about this but it's not a very easy to search term. Any pointers?
339.
by kaoD 2y ago
But Qwen is not multimodal, or is it?
340.
by kaoD 2y ago
So how do you document what is not in the code?
341.
by kaoD 2y ago
> ECS is like a whole another language on top of Rust, and it is dynamically typed. Can you elaborate? Or did you mean dynamically dispatched?
342.
by kaoD 2y ago
But that's yet-another-layer-of-abstraction with its own set of tradeoffs (e.g. I think CSS-in-JS is a trap, which seems to be the way for Lit; slots are still a thing; no SSR nor progressive enhancement; decorators!?!?!; etc.) which
343.
by kaoD 2y ago
Web components are the stuff that nightmares are made of. The amount of boilerplate I had to write just to keep DOM attributes and JS properties in sync was not fun, the impedance mismatch between them (DOM attributes being strings) was pai
344.
by kaoD 2y ago
In hopes of saving someone a search: UGC = User Generated Content.
345.
by kaoD 2y ago
Oh, I parsed that as "(is not ((divisible by three) or (is equal to three)))" and didn't think further.
346.
by kaoD 2y ago
> as long as the number of cards in it is not divisible by three or is equal to three I'm curious about the distinction here. Is it needed? Isn't N divisible by N? Or is there a mathematically-meaningful difference?
347.
by kaoD 2y ago
This is wrong. Deck is the set of cards that you bring to a match. Library is a region of the game state: the cards that are yet-to-be-drawn. This only exists as part of the gameplay. Other gamestate regions are the battlefield, the graveya
348.
by kaoD 2y ago
The price of non-conformity is higher -- e.g. they might ask you to explain why you didn't agree with the rest.
349.
by kaoD 2y ago
That doesn't explain why LLMs can't understand how many letters are in their tokens.
350.
by kaoD 2y ago
> There was no need to invent yet another operating system and desktop environment to replace the ones already existing. Actually there was. My web apps can be used in Windows, Linux, Mac, Android, iOS, Meta Horizon OS, etc. thanks to it
351.
by kaoD 2y ago
The semantics are indeed different in screen readers and the elements will be announced differently and behave differently in the accessibility tree (by virtue of having ARIA roles like e.g. "term"[0] or "definition"[1])
352.
by kaoD 2y ago
This is a very limited view that (understandably so) forgets that screen readers and other accessibility tools do offer more UX that is normally not obvious for us, able developers. This for example is very wrong if you take accessibility i
353.
by kaoD 2y ago
The more I'm reading this article the less I understand their point. Is there an actual paper that describes how their "random data" is generated? I can also generate random data that looks like any distribution by carefully
354.
by kaoD 2y ago
Ouch, $5b yearly of course!
355.
by kaoD 2y ago
Yep, that's exactly what I mean by "coding" as opposed to "engineering".
356.
by kaoD 2y ago
> AI is a very strange thing where 2 seemingly smart coders use it and one comes out thinking it's obviously revolutionary and the other one thinking it's a waste of time My take on this is that those 2 developers are often wor
357.
by kaoD 2y ago
They would need 250 million GPT Plus $20 subscribers to recoup a $5 billion expense. They're far from that even when we count the free users (which are likely 99% of the user base?) The math just doesn't work. They're hemorrh
358.
by kaoD 2y ago
I assume you don't want a full screen reader if you're not already using one, but if toasts are properly implemented (big if), screen readers can actually present them accessibly via the ARIA alert pattern[0]. Wanted to mention in
359.
by kaoD 2y ago
But can you make a decent job queue with anything that stores data? Not easily. E.g. you need atomicity if multiple consumers can take jobs, and I think you need CAS for that, not just any storage will do, right? You probably need ACI and
360.
by kaoD 2y ago
Honestly it's been a while for me too and I don't even know how to see the actual device partitions instead of just mounts (e.g. I think in this case your / is just a read-only ramdisk so it doesn't really matter what df

More ›