Hacker Newsnew | past | comments | ask | show | jobs | submit | more spcebar's commentslogin

How does people using it offset the amount of energy used to train it? If I use three hundred pounds of flour learning to make pizza, the subsequent three hundred pounds of flour I use making delicious pizzas doesn't make the first 300 go away. Am I misunderstanding the numbers?


Its not offset, its amortized. Your effective flour / pizza is (300 + 300) / num_pizzas. The total marginal flour expended will go up as you make more pizzas, but the effective cost will actually go down as the upfront cost is amortized over lifetime usage.


You don’t misunderstand the numbers, you misunderstand the point. If you flush your pizzas down the toilet, it’s a waste. If you feed 300 people with it, it’s not, even if you end up using the same amount of ingredients.


The cost of training has to be normalized by the number of users (or queries) that rely on that training.

If you use 300 lbs of flour to learn, and 300 lbs of flour to make 300 pizzas, then the total flour cost is 2 lbs of flour per pizza.


Sure, it's a value calculation.

If you're able to serve delicious pizzas afterwards, it was worth wasting the first kg (you might call it an investment).

If you're able to bring value to millions of users, it was worth to invest a few GWh into training.

You might disagree on the usefulness. I think, you shouldn't have wasted a kg of flour because I won't ever eat your pizzas anyway. But many (you, your guests, ChatGPT users) might think it was worth it.


It doesn't make it go away. Using your analogy - if you used 300lb to learn and then only made 10 lb of pizza after that, it would be a pretty poor use of resources.

If you instead went on to produce millions of pizzas for people and 30,000lb of flour, that 300lb you used to learn looks like a pretty reasonable investment.


Rick Adams, creator of Temple of ROM, runs the online museum for Colossal Cave: https://rickadams.org/adventure/ If you're interested in the history of the game, strongly recommend giving it a look.

Lovely, patient guy, who let me interview him when I was in college writing an essay on the game.


I had the exact same experience hosting my own text adventure on a Telnet server! Endless bots trying to log into the "press 1 to start, press 2 to load" screen. My secret dream was that a bot would write some weird sequence of commands and accidentally play through my game, though we never got there. Telnet is the lingua franca of insecure IoT devices, so it's not surprising (though sad) that any FTP server will immediately become a juicy target for bots. Wrote a blog post about it here, of anyone's interested. https://benergize.com/2020/09/06/some-people-hate-it-oldscho...


In theory a web developer is already overriding these styles, and if they're not, they can. Default browser styles are never to be taken for granted.


The fact that every webdev carries around/trades personal "reset" stylesheets to undo everything the browser does by default is insane. It really highlights the disconnect between W3C and reality.


I think the fact that every webdev feels the need to override the user's agent and impose their own idea of what size a H1 should be is, well, kind of imposing. I might want to make H1 be 50 point comic sans. This should not matter to the web developer.


I don't understand how one can design a website that would survive arbitrary style changes. I think that is unrealistic, so the designer should expect all default styles to be standard. And if the user has too much free time to change the font size then it is their own problem; my suggestion is that they simply use reader mode and change the styles there.

I would like to remind that some time ago browsers allowed to change the default font size; it never worked well so Opera started to scale the whole page instead. Other browsers followed it.

Android browsers seem to repeat the same mistake by the way: they override developer's styles when the user changes font size in OS accessibility settings.


I mean, it's my computer. I should ultimately be in control of how a document renders. I might need larger text due to poor eyesight, or need to use a screen reader. Or I might just be irritated that the web developer just up and decided that light gray on dark gray text looked cool (it doesn't, it barely can be read). Or I might want to scroll with my keyboard because a mouse is painful to my RSI. If I go out of the way and set up accessibility settings, I would expect all applications on my device (including the web browser) to respect those settings.


It matters because we’re the one that has to deal with ticket when user complains that their tweaked-to-within-an-inch-of-its life system inevitably breaks the world.

The problem isn’t you per say, it’s the 5000 people that mis-follow some YouTube video because it looks cool without it actually understanding what they’re changing , how to undo it, or what the implications are.


Mozilla doesn't agree with you; if they thought that was true they wouldn't do the weird phased rollout to stable.

Although I agree with codedokode insofar as I don't see how the phased rollout in stable could possibly help. Hopefully they've thought of something I haven't otherwise it is silly.


You shouldn't but there are many sites that assume the default font is black and the default background is white. (I'm sure I forgot to set them myself.)

There are [to] many ways to set the font size. I don't even know which on is the correct choice, if there is such a thing.

Maybe not trying to control it is the best approach? How can one tell?


One of my favorite books by my favorite author. Tracy Kidder gets to the heart of everyone he profiles like no other. Strongly recommend his book House if you like Soul of a New Machine.

Funny anecdote: I've read Soul of a new machine many times over the years, and love it dearly. I stayed with my aunt and uncle last year and somehow the topic of mini computers came up. I mentioned Soul of a New Machine and my uncle said "oh yeah I worked on that computer." I was blown away. Apparently he had worked at Data General and he was on the software team (which is briefly mentioned towards the end of the book) that wrote the software the Eagle was released with.


> // I dislike having a prefixed unit as the base reference. What a horrible decision. Why don't you just have it go to ten and make ten a little louder?

Really delightful.


That question (regarding the unit of mass, not the volume control) does apparently have an answer[1,2], and the answer (if true) is really funny: this was in fact the original plan, and the proposed unit was called the grave[3] (pronounced “grahv”). Then the French Revolution came and that name’s similarity to the word grave (count, the noble title, from German; cf. English margrave) led to it being superseded by the gram.

I don’t like the reference I have, though, so take this whole story with a grain of salt.

[1] https://phys.libretexts.org/Courses/Prince_Georges_Community...

[2] https://physics.stackexchange.com/questions/64562

[3] https://en.wikipedia.org/wiki/Grave_(unit)


Conversely, a story from the opposite end of the extreme: Fresh out of college I got an interview at a company that made games. It was exactly what I wanted to do and I made it miraculously far given the fact that I had zero real world experience. I made it to round three but ultimately got rejected. Who did they hire instead? A friend who works at the company later told me that it was down to two candidates. It was me vs the guy who wrote the game engine they make all their games in.


Original study: https://www.cell.com/current-biology/abstract/S0960-9822(24)...

It's worth noting that the chimps were in captivity, so it's not like there were researchers crouching in a bush in the jungle watching chimps pee for 600 hours. At least not for this study.


> Let’s be real: In effect, pronouns are a form of generational warfare.

Oh, **** off.

I agree that grace is something that sometimes gets lost in the cutting edge of liberalism. The article touches on this—in some cases, yes, it's a generational difficulty conceptualizing what are to them new ideas about gender that conflict with their upbringing. There's sometimes an attitude of "get with the program yesterday" amongst very progressive people that serves to alienate people acting in good faith trying but erring. I think if we did show more grace in that area, it would be a net positive for society and for trans acceptance.

However. That "get with it yesterday" attitude is a response to the overwhelming amount of bad faith dialogue that currently exists and has historically existed in the trans conversation. This article glosses over intentional misgendering and real transphobia in a way I found very gross.

Starting a conversation with pronouns can be a form of liberal virtue signalling, but it can also be a genuine exercise in trying to accommodate people's hard fought for identities. I'm sorry if you don't think someone is worth gendering correctly because they haven't done something outstanding for you, and I'm sorry if you don't feel that person is worth expressing your identity to either. To the same extent that we should be extending grace to people acting in good faith to understand gender pronouns, I hope we can extend that grace to people using them.


The technical term is babbling! It's a step on the road to acquiring language and it occurs in different phases at different ages for different infants.

https://en.m.wikipedia.org/wiki/Babbling


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: