Well, it seem damned easy for Amazon to just delete the fridges, so it being on two different continents really didn’t matter, did it?

I’m talking about the general sentiment from articles like this, not the article itself. The content of the article doesn’t really matter in the grand scheme of things.

It’s the Constant. Fucking. Beratement. of the technology.

Like, we fucking get it: You’re a technophobe and hate technology, and love to write articles that shit on LLMs, because that’s what gets clicks. And judging from the votes from this forum, most everybody falls for the clickbait, which then generates even more hateful articles because they know it gets them views.

Meanwhile, out there in the real world, people go to work, and use this sort of technology in their day-to-day jobs. There’s this extreme and jarring disconnect between public opinion, what the news report, and what’s actually happening in real life. I feel like I’m watching Fox News half the time. It’s like all of these haters of LLMs suffer from a massive cognitive dissonance when they are in the workplace. Or they are so behind the times that they aren’t using this technology. Or they don’t even realize the things they use are using this technology behind the scenes.

Exactly. There’s a finite amount of time available to teach somebody all of the useful skills needed to live life and build skills for a career.

Schools are no longer teaching cursive, or if they do, they don’t spend a lot of time on it. Same thing with all of the manual math operations. Learning algebra is more important. Hell, learning how to use a calculator is more important.

Calling it cheating is about as dumb as when math teachers called calculators cheating. If everybody has access to a calculator that can process any division math problem you throw at it, learning how to do long division is suddenly not very useful.

Imagine it’s the late 90s to early 2000s, and millions of people are on this anti-Internet bandwagon, while scores and scores of articles (on paper, of course) are always pushing this negative slant towards the Internet. People reading this shit about how the Internet is going to doom us all, and we should reject it in favor of traditional media and research.

This is what these last few years feel like. Just an outright rejection of useful and life-changing technology, while the corpos embrace it. The complete 180 to how the late 90s actually turned out, when corpos were slow on the uptick with this whole Internet thing.

“I put my carton of eggs in the fridge, and the fridge fell over, breaking all of my eggs.”

I blame copyright law for requiring this.

That’s not true. If you give it context, it understands and retains context quite well. The thing is that you can’t just say “write code for me” and expect it to work.

Also, certain models are better than certain tasks than others.

Yep, and the general public is too stubborn to accept a little thing like nuance. Neither are the CEO assholes that can’t stop talking about layoffs and replacing jobs, out in the open.

State monopolies are fine. The government is owned by the people, and when corporations don’t think an industry is profitable, the government’s job is to step it and take on that job themselves. That’s how we end up with utilities, 911 services, the post office (back when it wasn’t fucked with by Republicans).

The problem with copyrights is the corporate stank that gatekeeps enforcement. When a large corporate entity sues a small party, the small party is fucked.

Also, Mark Twain and Disney fucked up the length of copyright over the last 150 years. The social agreement was that we were supposed to get most of this shit into public domain in a reasonable amount of time. 80 years + life of the author is not a fucking reasonable amount of time, by any stretch of the imagination.

Copyrights exists to maintain profits of large corporations. Copyright, patents, and intellectual rights were created under the false pretense that it “protects the little person”, but these are lies told by the rich and powerful to keep themselves rich and powerful. Time and time again, we have seen how broken the patent system is, how it is impossible to not step on musical copyright, how Disney has extended copyrights to forever, and how the megacorporations have way more money than everybody else to defend those copyrights and patents. These people are not your friend, and their legal protections are not for you.

So, fuck them. Let AI destroy the fabric of copyrights.

It’s also a dumb fucking argument in general.

LLM models are not magic. They don’t store the entirety of human history in a 10-20GB file. That’s a fucking stupid idea. They look at something, process some weights, and move on to the next thing.

It’s like going on Google Image Search, and seeing all of the copyrighted images on your browser. You didn’t “steal” those, just because it was downloaded. That’s not how copyright works.

Please stop parroting this stupid fucking argument.

Blame Melinda Tankard Reist. She started this whole mess.

She could have, but she didn’t, which is why she got her current job.

I completely forgot that Holden was part of that commune of, what, eight mothers?

It’s not really “your house connected to the Internet” as much as just connecting it together to a central hub. My lights dim and turn off at set times of the day using scenes, including a more dynamic sunrise/sunset time. If it’s stormy weather during the day, the front porch light comes on. I have a few colored lights that change colors with the seasons. Home Assistant brings down the garage door at night, if I forget. If the garage sensor is tripped, the garage light comes on. I have dumber motion sensors for locations where it makes sense, like pantries or closets.

I can control all of my interfaces from my phone. It can hook into local cameras, so that everything is centralized. HA can hook into any of the voice assistant systems, like Alexa. There are so many little conditions and automated switches I could create, based on either outside stimuli (like the weather) or internal sensors.

It’s a shame, because ComfyUI can be so much more than just image generation. And just because there’s a lot of string processing for LLMs doesn’t mean that it isn’t important to capture in an I/O interface, especially when it comes to preserving chat history. Save data, load data, ask new questions, etc.

ChatGPT is pretty damn powerful, I’ll admit. But, all of its components need to be localized, especially since something like a Mixture of Experts model could be split down to base models and loaded/unloaded as necessary.

The Home Assistant community has a good reputation of routing around that kind of bullshit. There’s enough users that I predict a homegrown module will pop up in the next few months.

I love my Ratgdo garage door unit, routing around the bullshit proprietary LiftMaster APIs that want you to download some stupid phone app. No, fuck you. I hard-wire that shit into my garage door opener and get an API that works like a dream.

Sometimes, trying to integrate smart home devices can be a pain, but the Home Assistant core makes the integration much much easier. Please don’t like shit like this to detract from a future smart home project for your house. I would encourage you to just start it, if you’re wanting to get into it and willing to learn some things.

ComfyUI is just a bunch of Python code tied into I/O nodes. I’m surprised there isn’t a good set of nodes for SGLang yet.

Have you used any good ComfyUI workflows specifically for chat LLMs?