Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GPT-4 details leaked?

threadreaderapp.com

Post reply on HN
Print view View original

121–130 of 648 posts

Re: GPT-4 details leaked?

#121

Post by flir »

downWidOutaFite wrote: post #85
PostOnce wrote: post #38

Earlier quoted context omitted.

It's several reasons, first it's the lies and the abuse of a charity, that wouldn't be an issue if they had began as a private company instead of robbing a charity. But secondly, even if they were a private company, it's dishonest and reprehensible to claim to congress that you want to "protect the public" when you really only give a shit about protecting your moat, I'm not happy about that either. I'm also tired of…

The "regulatory capture" conspiracy theory makes no sense to me. It takes 9 figures in cash to create and run one of these super big models. Only big tech was ever going to create them, and big tech is already very experienced at navigating regulation, regulation wasn't ever going to stop them from competing. And in general, our democracy works better than the nihilist libertarians give it credit for.

I bet it'll be 6 figures within 18 months.

Re: GPT-4 details leaked?

#122

Post by lacrimacida »

jackstraw14 wrote: post #111
vouaobrasil wrote: post #95

Earlier quoted context omitted.

I would love to see your rebuttals, especially since I have never seen any strong arguments in favour of AI being a net benefit to society, and I have thought and read about this at lenght. Of course, I always expect downvotes on my posts here since there is a strong tendency towards loving technology here. But what I find most interesting is that there is absolutely no taking of responsibility of any technological c…

> But what I find most interesting is that there is absolutely no taking of responsibility of any technological creations. I appreciate your willingness to talk about it, but to be honest it doesn't seem like it matters much what you, or any of us (not singling you out in particular), thinks about it, does it? It probably doesn't even matter who these people are who should take responsibility. This is one genie, like…

It actually does make a difference. The genie is out of the bottle partially but it depends a lot on what we allow it it be used on. If we sit idly and allow for ingesting all what’s written for instance, including whats currently written and let bros make derivative works for a quick buck then we mostly killed the writer’s incentive to write or publish. If we slow down and not allow ripping one another off it could have the opposite effect and trully lift all the boats at once. There’s a concurent thread about Sarah Silverman suing OpenAI, that’s what Im talking about as well here

Re: GPT-4 details leaked?

#123

Post by Vecr »

Llamamoe wrote: post #103
juliensalinas wrote: post #100

Earlier quoted context omitted.

LLaMA 30B or 60B can be very impressive when correctly prompted. Deploying the 60B version is a challenge though and you might need to apply 4-bit quantization with something like https://github.com/PanQiWei/AutoGPTQ or https://github.com/qwopqwop200/GPTQ-for-LLaMa . Then you can improve the inference speed by using https://github.com/turboderp/exllama . If you prefer to use an "instruct" model à la ChatGPT (i.e. tha…

How low can you get the memory and computational power requirements that way?

You can run that model (Wizard-30) on a computer with 64 gigabytes of RAM (or smaller, I don't know how tight you can cut it). You obviously want fast RAM and a good CPU, but you don't need a GPU.

Re: GPT-4 details leaked?

#124

Post by aussieguy1234 »

The fact they are using MoE is interesting. There are alot of specialised open source models on HuggingFace. You just need an LLM to act as the core "brain" and a few other components.

HuggingGPT works similar to this. It automatically chooses, downloads and runs the right "expert" model from HuggingFace https://arxiv.org/abs/2303.17580

Re: GPT-4 details leaked?

#125

Post by TeMPOraL »

BelleOfTheBall wrote: post #119
ly3xqhl8g9 wrote: post #115

Earlier quoted context omitted.

From the post: [Rumors that start to become lawsuits] Some speculations are: - LibGen (4M+ books) - Sci-Hub (80M+ papers) - All of GitHub This is the most funny, but in the end sad aspect. If ChatGPT was indeed trained on pirated content and is able to be(come) such a powerful tool, then the copyright laws should have been abolished yesterday. If ChatGPT was not trained on all these resources out there, then think ho…

That is a very dangerous way to approach copyright laws. They are definitely abused by corporations like Disney, infamously so, but abolishing them is absolutely not the answer. Art makers are already struggling en masse, taking away their ability to earn money off their work isn't an answer, especially if it's just to train a predictive text generator.

Counterpoint: there's way too much "art makers", copyright is keeping the quality ceiling down, good content loses monetization game with spam - all while that "predictive text generator" is, for better or worse, pretty much the most magnificent piece of technology invented this century, and - for better or worse - it's likely to become next major shift to economy and life.

Re: GPT-4 details leaked?

#126

Post by ben_w »

flir wrote: post #121
downWidOutaFite wrote: post #85

Earlier quoted context omitted.

The "regulatory capture" conspiracy theory makes no sense to me. It takes 9 figures in cash to create and run one of these super big models. Only big tech was ever going to create them, and big tech is already very experienced at navigating regulation, regulation wasn't ever going to stop them from competing. And in general, our democracy works better than the nihilist libertarians give it credit for.

I bet it'll be 6 figures within 18 months.

Even if it was free to train, FAANGs can beat OpenAI on spending to follow regulations.

Re: GPT-4 details leaked?

#127

Post by TeMPOraL »

dclowd9901 wrote: post #106
maximamas wrote: post #101

Earlier quoted context omitted.

It’s ironic you say: “we are playing with fire.” Playing with fire is, in large part, literally how humans have come to dominate this planet. Why stop now?

To turn your metaphor on its head, we aren’t playing with fire when we use it constructively; rather we are very carefully and thoughtfully deploying it, no doubt due to our gradual and deadly lessons with it over time. When we “play” with it (a la fireworks or neglected campfires), it wreaks rampant destruction. Being we are basically toddlers with this new technology, I would argue the breathless speed at which it’…

Counterpoint: "playing with it" is the only way we have to actually master something. "Carefully and thoughtfully deploying it" only comes way after many people first extensively played with it (for any specific "it"), first because of curiosity (i.e. for shits and giggles), then for a quick buck.

Re: GPT-4 details leaked?

#128

Post by LiamPowell »

rjb7731 wrote: post #110

I've previously noticed when playing with GPT-4 it can sometimes 'autocomplete' on different sections of the text its feeding back, sometimes what looks like 4 or more different sections. Might be unrelated but is this MoE in action or them streaming the response in some way?

This is just an issue with their frontend that seems to occur when it encounters \n\n. The actual data coming in only changes at the end of the message.

Re: GPT-4 details leaked?

#129

Post by Al0neStar »

Roark66 wrote: post #82

The tweet is gone. What was in it? Also, I'm dubious about this unsubstantiated claim. The biggest past innovation (training with human feedback) actually shrunk the size of a model. Compare Bloom-366B with falcon-40B (much better). I would be mildly surprised if it turned out Gpt4 has 1.8T parameters. (even if it's a composite model as they say) The article says they use 16 experts 111B each. So the best thing to as…

Maybe 111B is the base GPT-3.5 model.

Re: GPT-4 details leaked?

#130

Post by xeckr »

mmahemoff wrote: post #96

I've been wondering how freemium services like Thread Reader still operate now that Twitter is charging prohibitive prices for API access and taking measures to prevent scraping. The cheapest API plan with read access is $100/month, which reads 10,000 tweets, so could only produce about 500 pages like this one on demand.

const puppeteer = require('puppeteer'); and so on and so forth.
Post reply on HN