Live data from Hacker News

Meta Llama 3

llama.meta.com

671–680 of 965 posts

Re: Meta Llama 3

#671

Earlier quoted context omitted.

It's a blob that costs over $10,000,000 in electricity costs to compile. Even if they released everything only the rich could push go.

There is an argument to be made about the importance of archeological preservation of the provenance of models, especially the first few important LLMs, for study by future generations. In general, software rot is a huge issue, and many projects which may be of future archeological importance are increasingly non-reproducible as dependencies are often not vendored and checked into source, but instead downloaded at co…

This is comment is cooler than my Arctic Vault badge on GitHub.

Who were the countless unknown contemporaries of Giotto and Cimabue? Of Da Vinci and Michelangelo? Most of what we know about Renaissance art comes from 1 guy - Giorgio Vasari. We have more diverse information about ancient Egypt than the much more recent Italian Renaissance because of, essentially, better preservation techniques.

Compliance, interoperability, and publishing platforms for all this work (HuggingFace, Ollama, GitHub, HN) are our cathedrals and clay tablets. Who knows what works will fill the museums of tomorrow.

Re: Meta Llama 3

#672

Earlier quoted context omitted.

Wild considering, GPT-4 is 1.8T.

The original GPT4 may have been around that size (16x 110B). But it's pretty clear GPT4 Turbo is a smaller and heavily quantized model.

Yeah, it’s not even close to doing inference on 1.8T weights for turbo queries.

Re: Meta Llama 3

#673

Earlier quoted context omitted.

> I just want to express how grateful I am that Zuck Praise for him at HN? It should be enough of a reason for him to pop a champagne today

Yeah, I'm also surprised at how many positive comments are in this thread. I do hate Facebook, but I also love engineers, so I'm not sure how to feel about this one.

> I do hate Facebook, but I also love engineers, so I'm not sure how to feel about this one.

"it's complicated". Remember that? :)

It's also a great way to avoid many classes of bias. One shouldn't aspire to "feel" in any one way. Embrace the complexity.

Re: Meta Llama 3

#674
post #316

Earlier quoted context omitted.

> It's fun to be able to retire early or whatever, but driving software engineer salaries out of reach of otherwise profitable, sustainable businesses is not a good thing. That argument could apply to anyone who pays anyone well. Driving up market pay for workers via competition for their labour is exactly how we get progress for workers. (And by 'treat well', I mean the whole package. Fortunately, or unfortunately,…

Bubbles are bubbles. They can be enjoyed/exploited (early retirment, savvy caching of excess income, etc) by workers but they don't win anybody progress and aren't a thing to celebrate. Workers (and society) have not won progress when only a handful of companies have books that can actually support their inflated pay, and the remainder are ultimately funded by investors hoping to see those same companies slurp them u…

> They can be enjoyed/exploited (early retirment, savvy caching of excess income, etc) by workers but they don't win anybody progress and aren't a thing to celebrate.

Huh, if I get paid lots as a worker, I don't care whether the company goes belly up later. Why should I? (I include equity in the total pay package under judgement here, and by 'lots' I mean that the sum of equity and cash is big. If the cash portion is large enough, I don't care if the stock goes to zero. In any case, I sell any company stock as soon as I can, and invest the money in diversified index funds.)

> Workers (and society) have not won progress when only a handful of companies have books that can actually support their inflated pay, and the remainder are ultimately funded by investors hoping to see those same companies slurp them up before the bubble bursts.

I'm more than ok with willing investors (potentially) losing capital they put at risk. Just don't put some captive public retirement fund or task payer money into this. Those investors are grown up and rich, they don't need us to know better what is good for them.

> Workers don't win progress when they're lured into then converting that income into impractical home loans that bind the workers with golden handcuffs and darkly shadow their future when the bubble bursts.

This says more about carefully managing the maximum amount of leverage you want to take on in your life. It's hardy an argument that would convince me that lower pay is better for me.

People freak out when thinking about putting leverage in their stock portfolio, but they take on a mortgage on a house without thinking twice. Even though getting out of a well diversified stock portfolio and remove all the leverage takes less than half an hour these days (thanks to online brokers), but selling your single concentrated illiquid house can take months and multiple percentage points of transaction costs (agents, taxes, etc).

Just don't buy a house, or at least buy within your means. And make sure you are thinking ahead of time how to get out of that investment, in case things turn sour.

> Workers win progress when they can practice their trade with respect and freedom and can and secure a stable, secure future for themselves and their families.

Guess who's in a good negotiation position to demand respect and freedom and stability from their (prospective) employer? Someone who has other lucrative offers. Money is one part of compensation, freedom and respect (and even fun!) are others.

Your alternative offers don't all have to offer these parts of the package in the same proportions. You can use a rich offer with lots of money from place A, to try and get more freedom (at a lower pay) from place B.

Though I find that in practice that the places that are valuing me enough to pay me a lot, also tend to value me enough to give me more respect and freedom. (It's far from a perfect correlation, of course.)

> Software engineers didn't need these bubble-inflated salaries to acheive that.

Yes, have lived on a pittance before, and survived. I don't strictly 'need' the money. But I still firmly believe that all else being equal that 'more money = more better'.

> What fight we do still need to make is on securing non-monetary worker's rights and professional deference, [...].

I'd rather take the money, thank you.

If you want to fight, please go ahead, but don't speak for me.

And the whole thing smells a lot like you'd (probably?) want to introduce some kind of mandatory licensing and certificates, like they have in other engineering disciplines. No thank you. Programming is one of the few well paid white collar jobs left where you don't need a degree to enter. Let's keep it that way.

Re: Meta Llama 3

#675
post #513

Earlier quoted context omitted.

Love how everyone is romanticizing his engineering mindset. But have we already forgotten that he was even more passionate about the metaverse which, as far as I can tell, was a 50B failure?

If you actually listen to how Zuck defines the metaverse, it's not Horizons or even a VR headset. That's what pundits say, most of whom love pointing out big failures more than they like thinking deeply. He sees the metaverse as the entire shared online space that evolves into a more multi-user collaborative model with more human-centric input/output devices than a computer and phone. It includes co-presence, mixed r…

50 billion dollars and fewer than 10 million MAU. That's a massive failure.

Re: Meta Llama 3

#676
post #316

Earlier quoted context omitted.

> It's fun to be able to retire early or whatever, but driving software engineer salaries out of reach of otherwise profitable, sustainable businesses is not a good thing. That argument could apply to anyone who pays anyone well. Driving up market pay for workers via competition for their labour is exactly how we get progress for workers. (And by 'treat well', I mean the whole package. Fortunately, or unfortunately,…

> Driving up market pay for workers via competition for their labour is exactly how we get progress for workers. There's a difference between "paying higher salaries in fair competition for talents" and "buying people to let them rot to make sure they don't work for competition". It's the same as "lowering prices to the benefit of consumer" vs "price dumping to become a monopoly". Facebook never did it at scale thoug…

> It's the same as "lowering prices to the benefit of consumer" vs "price dumping to become a monopoly".

Where has that ever worked? Predatory pricing is highly unlikely.

See eg https://www.econlib.org/library/Columns/y2017/Hendersonpreda... and https://www.econlib.org/archives/2014/03/public_schoolin.htm...

> Facebook never did it at scale though. Google did.

Please provide some examples.

> There's a difference between "paying higher salaries in fair competition for talents" and "buying people to let them rot to make sure they don't work for competition".

It's up to the workers themselves to decide whether that's a good deal.

And I'm not sure why as a worker you would decide to rot? If someone pays me a lot to put in a token effort, just so I don't work for the competition, I might happily take that over and practice my trumpet playing while 'working from home'.

I can also take that offer and shop it around. Perhaps someone else has actual interesting work, and comparable pay.

Re: Meta Llama 3

#677

Earlier quoted context omitted.

Love how everyone is romanticizing his engineering mindset. But have we already forgotten that he was even more passionate about the metaverse which, as far as I can tell, was a 50B failure?

It's a bit too early IMHO to declare the metaverse a failure. But that said, I don't think it matters. I don't know anybody who hasn't been wrong about something , or made a bad bet at times. Even if he is wrong about everything else (which he's not, because plenty of important open source has come out of facebook), that doesn't change the extreme importance that is Llama and Meta's willingness to open things up. It'…

$50B for <10M MAU is absolutely a failure, today, as I'm typing this.

Re: Meta Llama 3

#678

Earlier quoted context omitted.

Love how everyone is romanticizing his engineering mindset. But have we already forgotten that he was even more passionate about the metaverse which, as far as I can tell, was a 50B failure?

The Quest is the top selling VR headset by a very large margin. He's well positioned to take that market when it eventually matures a bit. Once the tech gets there, say in a decade we might see most people primarily consume content via VR and phones. That's movies, games, TV, sporting events, concerts.

Give me $50 billion dollars and I'll bet I could get 8 million MAU on a headset. It's a massive failure because Zuck's a nerd and not a product guy.

Re: Meta Llama 3

#679

Earlier quoted context omitted.

Love how everyone is romanticizing his engineering mindset. But have we already forgotten that he was even more passionate about the metaverse which, as far as I can tell, was a 50B failure?

was a failure? they are still building it, when they shut down or sell off the division then you can call it a failure

10 years, $50 billion, fewer than 10 million MAU. It's a failure today, right this minute it's a failure.

Re: Meta Llama 3

#680

Earlier quoted context omitted.

It's likely RAG / augmented with web data. Would be interested if local execution returned the same results.

It certainly doesn't, I'm running the 7B locally with ollama It provided a lot more detail about the case, but does not have current information. It hallucinated the question about juror count, or maybe confused it with a different case seems more likely, one of the E Jean Carroll cases or the SDNY Trump Org financial fraud case?

You: how many jurists have been selected in the Trump trial in New York?

Meta AI: A full jury of 12 people has been selected for former President Donald Trump's trial in New York City, in addition to one alternate ¹. The selection process will continue in order to select five more alternates, though it is hoped that the selection process will be finished tomorrow ². Once all alternates have been selected, opening statements can begin ¹.

--- https://www.nbcnews.com/politics/donald-trump/jury-selection... https://www.cbsnews.com/news/donald-trump-new-york-court-cri...

Post reply on HN