Skip Navigation

User banner

Poik

@ Poik @pawb.social

Posts
0
Comments
98
Joined
3 yr. ago

  • Very much smart people

    Jump

  • ... 1957

    Perceptrons. The math dates back to the 40s, but '57 marks the first artificial neural network.

    Also 35 years is infancy in science, or at least teenage, as we see from deep learning's growing pains right now. Visualizations of neural network responses and reverse engineering neural networks to understand how they tick predate 2010 at least. Deep Dream was actually built off an idea of network inversion visualizations, and that's ten years old now.

  • Very much smart people

    Jump

  • When you're working at the algorithm level, you get funny looks... Even if it gets to state of the art results, who cares because you can throw more electricity and data at it instead.

    I worked specifically on low data algorithms, so my work was particularly frowned upon by modern ai scientists.

    I'm not doxxing myself, but unpublished work of mine got published in parallel as Prototypical Networks in 2017. And everyone laughed (<- exaggeration) at me researching RBFs which were considered defunct. (I still think they're an untapped optimization.)

  • Boromir, his hand in an O

    Jump

  • Can a Brit confirm this?

    Jump

  • So called because the toast in military kitchens were nicknamed shingles, as in roofing tiles. Evocative of bad cooking, which I'm betting was rampant.

    Honestly, shit on a shingle (s.o.s. appropriately) is better than it sounds, even when not referred to under that name. But it's definitely a comfort food. It's not good for you, it's just creamy, beefy, and starch. Inoffensive, cheap, and easy to make in bulk. (Kinda want some now.)

  • Defensive much?

    Jump

  • I love this analogy. May I steal it?

  • Gets confusing

    Jump

  • In the US, in most states, getting caught or recognized is enough to put you on the sex offender list. Even if you're in private. (Again, in most states.) And that means you can no longer move into a new home without informing all your neighbors that you're a sex offender for the rest of your life, among other penalties. There's no difference to the US between this and people who actually do sexual crimes when it comes to this punishment.

  • What Are People Still Doing on X?

    Jump

  • Trump kills rural broadband grants, calls digital equity program “racist and illegal”

    Jump

  • The problem isn't the name. The problem is Trump already hates the letters DEI and the word equity. He's declared war on anything that attempts to dismantle institutionalized bigotry. And he doesn't recognize that rural equity isn't about race, hence him declaring it racist.

  • Sometimes we create our own problems.

    Jump

  • Or use it on large scale computing for protein folding simulations, or something.

    And yeah, gravity batteries is the best I think we have, with water being the most common medium with pumped-storage hydroelectricity. But the scales of the things are kind of incongruent and... Autoincorrect actually got it right trying to correct that to inconvenient. Still really cool. I think we may need some innovations to cut down on scale issues though. Although it looks like the total power storage available is about one day worth of power for the US in PSH, I'm curious if the instantaneous output is sufficient for the grid and how spread out the storage locations are, as I somewhat doubt they're often in flatter regions. All in all, I'm not a power engineer, I just know a few and I should bug them sometime.

  • Sometimes we create our own problems.

    Jump

  • We didn't really have good batteries at that scale. I believe the large scale power storage is still done using water and gravity. Which is honestly pretty neat, but requires lots of land and a high location.

  • You must choose...

    Jump

  • I wish I could taste the other flavors. That's actually why I was considering desensitizing myself to it. I get the dirt and earthy, but I love both of those the same. I've been growing my palette, but it took me nearly two decades to find hops that I could stand to start desensitizing myself to that bitter.

  • You must choose...

    Jump

  • I mean... It's literally genetic. The aldehydes in cilantro usually aren't strong enough for people to taste. But if you want to know what I taste when I eat cilantro, go crush a stink bug, it's the same chemical.

    Apparently I can desensitize myself to it, and I want to. Certainly would open up a lot of options in foods I'm already a fan of (if you leave out the cilantro).

  • Show us your cat's dating app profile pic

    Jump

  • Cuddlr

  • Călin Georgescu does a nazi salute after leaving the General Prosecutor's office (Romania)

    Jump

  • Sorry, I saw Nazis in the parent post and didn't realize they meant neo-nazis. I was addressing their claim as written, not within the current context. And no, I don't think they should have to differentiate, a Nazi is a Nazi. I just don't pay that much attention sometimes.

  • Călin Georgescu does a nazi salute after leaving the General Prosecutor's office (Romania)

    Jump

  • Russia was an ally of Germany via their non-aggression pact of 1939 until Germany tried to invade them and ended the alliance in 1941. Not a very long lived alliance, but it tells a little more of the story.

  • About as open source as a binary blob without the training data

    Jump

  • Hm. I speak like a bot, do I? Maybe I am autistic after all.

    I am aware, my boyfriend and I have already had this conversation, but I guess he's not on Lemmy, so you can't ask him.

    Yes, DeepSeek caused a drop in the stock price, but you were saying that believing that LLM's are over-hyped would lead to having insider knowledge and could give us an advantage in the stock market. Particularly with their already tanked stock. However, the stock market fluctuates based on hype, not value, and will do whatever the fuck it pleases, so the only way to have insider knowledge is by being on a board who controls the price or by managing to dump hype into the system. That is not something a lot of people have the power to do individually.

    But since you think I'm a bot and I have no way to disprove that thanks to what the world is now, I bid you adieu. I hope you're having a good one. And stop antagonizing people for talking differently, please.

    Edit: I took a look at your recent comment history, and you do come off as trying to troll and be disingenuous. If you want to have a less inflammatory conversation, you can DM me, but I do recommend you tone it down. You're not helping anyone with how you're approaching this, buddy.

  • About as open source as a binary blob without the training data

    Jump

  • I'm glad we agree. I don't know how much mental energy we should devote to these things, but I guess I'm happy to see discussion on this platform. I kind of miss the days when I had to respond to people who watched I, Robot and think I'm trying to destroy humanity, instead of... I don't know what to call this except basically the same thing they did in the 60's before the first AI winter, but with more drastic consequences.

  • About as open source as a binary blob without the training data

    Jump

  • But also, you were talking about Nvidia in your comment I responded to, not Deepseek, so your rebuttal is non sequitur...

  • About as open source as a binary blob without the training data

    Jump

  • Actually no. As someone who prefers academic work, I very heavily prefer Deepseek to OpenAI. But neither are open. They have open weights and open source interpreters, but datasets need to be documented. If it's not reproducible, it's not open source. At least in my eyes. And without training data, or details on how to collect it, it isn't reproducible.

    You're right. I don't like big tech. I want to do research without being accused of trying to destroy the world again.

    And how is Deepseek over-hyped? It's an LLM. LLM's cannot reason, but they're very good at producing statistically likely language generation which can sound like its training data enough to gaslight, but not actually develop. They're great tools, but the application is wrong. Multi domain systems that use expert systems with LLM front ends to provide easy to interpret results is a much better way to do things, and Deepseek may help people creating expert systems (whether AI or not) make better front ends. This is in fact huge. But it's not the silver bullet tech bros and popsci mags think it is.

  • About as open source as a binary blob without the training data

    Jump

  • ... Statistical engines are older than personal computers, with the first statistical package developed in 1957. And AI professionals would have called them trained models. The interpreter is code, the weights are not. We have had terms for these things for ages.