Skip Navigation

Rhaedas

@ Rhaedas @fedia.io

Posts
1
Comments
1786
Joined
3 yr. ago

There will be times when the struggle seems impossible. Alone, unsure, dwarfed by the scale of the enemy. Freedom is a pure idea. It occurs spontaneously and without instruction. Random acts of insurrection are occurring constantly. There are whole armies, battalions that have no idea that they've already enlisted in the cause. And even the smallest act of insurrection pushes our lines forward. Tyranny requires constant effort. It breaks, it leaks. Authority is brittle. Oppression is the mask of fear.

  • Removed

    4 Tesla owners react to their FSD dreams getting crushed: 'It feels like a bait and switch'

    Jump

  • Yes, if you read them. If you consider doing a Hail Mary in bad weather and managing to not hit anything better, then I guess they are better... at taking risks.

  • Removed

    4 Tesla owners react to their FSD dreams getting crushed: 'It feels like a bait and switch'

    Jump

  • You're correct on AI. But I laughed at you saying humans are predictable. Seen any dashcam footage?

  • Removed

    4 Tesla owners react to their FSD dreams getting crushed: 'It feels like a bait and switch'

    Jump

  • Humans aren't better at driving in bad situations, they're just better at ignoring most of the input and focusing on one thing, and more importantly, taking risks, which a computer isn't going to be programmed to do. If a human navigates through a bad rainstorm, barely able to see anything, and makes it out fine, then they claim they're better than a self-driving car which would shut down. Or more simply, a route that is very tight and risky, but a human will YOLO and make it through. They're not better, they're just lucky a lot.

  • "You guys seeing this?!"

    Jump

  • One of the better episodes. Seemed a bit goofy at that point, but his breakdown right after was scary.

  • Trump says it’s ‘treasonous’ to say US not winning war in Iran

    Jump

  • We need a Tywin to send him to bed. He looks tired.

  • Bernie Sanders: AI oligarchs do not want to just replace specific jobs, they want to replace the working class. We must fight back.

    Jump

  • If we take that approach with any change, that it might be possible but it also could fail or become corrupt, then we're just going to spin wheels. Like we're doing. We have to change something, this isn't working.

  • Bernie Sanders: AI oligarchs do not want to just replace specific jobs, they want to replace the working class. We must fight back.

    Jump

  • UBI isn't viable without other changes, but it's just an improvement of various welfare systems, just without all the paperwork and exceptions. "Too expensive" is usually the excuse thrown around, which is funny given the news of how much killing or disappearing people is costing, all because some person in the White House needs a distraction.

  • Spotify adds 'Verified' badges to distinguish human artists from AI

    Jump

  • See, that's how you make actual users buy the badges. Now to be considered not AI you have to put money up. Who wins? Spotify does. And they don't even have to make verification all that thorough, just enough to say they tried to filter out the worst of the bots.

  • AI may be coming to Windows 11’s Clock app as Microsoft turns it into a focus tool

    Jump

  • "Lovely plumage"

  • Uncensored Models Actually Uncensored?

    Jump

  • It is certainly inaccurate, but in my mind's picture of how the transformers work, reducing their quantization and also doing what abliteration does, there is a line where you've done a lot of "damage" to the original model and so there will be places where it just hangs or goes off on severe tangents. There are good places for even the 1bit models where they don't get pushed to hard, but there are limits for them all, including the big ones.

    Hugging Face does have a few Q4_K_M versions. Maybe something will fit.

  • Deleted

    Permanently Deleted

    Jump

  • AGI might use LLM tech in their process, but LLMs by themselves aren't going to become aware. What happened is LLM tech became a gold mine, some who were doing AGI research jumped on it instead, and others followed. There is certainly still AGI research going on somewhere, but it's buried by the race to... something. The biggest problem I see, outside of the need for profit guiding all this, is that what they are building has become so complex they don't really understand it fully, they just keep finding ways to tack on things to get to some higher level without knowing why it works (or why it will break).

    And while LLMs aren't AGI, they still have the issue of misalignment, even without a self-awareness. We've seen early on the misdirection to obtain a goal, and the models now are more sophisticated. Maybe it's not their own goal, but a misunderstood goal that they'll say and do anything to get to.

    Good thing we're not putting them in control of important things, or full access to systems, right? Right?

  • Uncensored Models Actually Uncensored?

    Jump

  • It's as good as an 8B can be, but with the right system prompt for your purpose and proper expectations, I think it's good. I've had some other newer 8B that blew up after a few cycles, literally getting stuck on something, but I can't say this one ever did. But again, even the big models like Claude and the rest work better with short sessions and a specific, detailed prompt to start with. Use a model to make the prompt, telling it to be detailed, concise, and minimize fluff. Less tokens in and out that way, less context drift (hopefully).

  • Uncensored Models Actually Uncensored?

    Jump

  • Abliteration techniques might be more limited with reasoning models. I don't know if they process simply be rehashing the arguments or if there's more under the hood that would be harder to alter.

    I try new models from time to time, including some of the thinking ones, but I've always come back to the NeuralDaredevil model, even though it's "old". Your results may differ depending on the subject matter, but I can't think of an instance where I hit a wall. At most, maybe some sidetracking but once I told it to be more open it didn't hold back.

    I'm not sure what the appeal of the thinking mode is. Perhaps on some things it does better, but in watching its reasoning I've seen it talk itself out of a good solution too. Which is what you get with typical models when you push the context too far and don't start a new session, they wander.

  • Am I wrong?

    Jump

  • The other side of the hype is the graphical AI, diffusion models and related. Thanks to marketing for the LLMs, it's all been shoved into the AI label and blurred together. Not all AI is bad, and I would say the bad part is how the tool is being pushed towards use anywhere and everywhere, when it's only best for certain applications. The joke was always that when you get a hammer, everything looks like a nail, and yet we're doing that now.

  • Nothing can run Crysis (and live to tell the tale)

    Jump

  • Once they get them all finished and connected. And then only briefly.

  • U.S. war in Iran has cost $25B so far, says Pentagon official

    Jump

  • That's a lot of money that could have gone... no, wait, spending money to help people is too expensive.

  • The key to immortality

    Jump

  • Don't Look Up.

    Remember who escaped?

    The good news is, remember what happened to them. Maybe there is karma somehow.

  • US to issue passports featuring Trump’s picture to commemorate America’s 250th anniversary | CNN Politics

    Jump

  • What a petty little man. Got to put his name on everything because he knows deep inside he's worthless.

  • Writing advice

    Jump

  • However sometimes it's better to put it down for a while and go to something else. Just like with anything, you can get burned out and fixated on details, redoing things over and over in a circle.