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

Meta Llama 3

llama.meta.com

Post reply on HN
Print view View original

521–530 of 965 posts

Re: Meta Llama 3

#521

Post by jiggawatts »

mrtranscendence wrote: post #283
lordswork wrote: post #132

Earlier quoted context omitted.

Lex walked so that Dwarkesh could run. He runs the best AI podcast around right now, by a long shot.

I don't know Dwarkesh but I despise Lex Fridman. I don't know how a man that lacks the barest modicum of charisma has propelled himself to helming a high-profile, successful podcast. It's not like he tends to express interesting or original thoughts to make up for his paucity of presence. It's bizarre. Maybe I'll check out Dwarkesh, but even seeing him mentioned him in the same breath as Fridman gives me pause ...

He’s popular because of the monochrome suit, etc…

I don’t listen to a three hour interview to listen to the interviewer! I want to hear what the guest has to say.

Until now, this format basically didn’t exist. The host was the star, the guest was just a prop to be wheeled out for a ten second soundbite.

Nowhere else in the world do you get to hear thought leaders talk unscripted for hours about the things that excite them the most.

Lex enables that.

He’s like David Attenborough, who’s also worn the exact same khakis and blue shirt for decades. He’s not the star either: the wildlife is.

Re: Meta Llama 3

#522

Post by nojs »

modeless wrote: post #237
bbig wrote: post #16

Zuck has an interview out for it as well, https://twitter.com/dwarkesh_sp/status/1780990840179187715

Seems like a year or two of MMA has done way more for his charisma than whatever media training he's done over the years. He's a lot more natural in interviews now.

Alternatively, he’s completely relaxed here because he knows what he’s doing is genuinely good and people will support it. That’s gotta be a lot less stressful than, say, a senate hearing.

Re: Meta Llama 3

#523

Post by jug »

swalsh wrote: post #279
eigenvalue wrote: post #231

I just want to express how grateful I am that Zuck and Yann and the rest of the Meta team have adopted an open approach and are sharing the model weights, the tokenizer, information about the training data, etc. They, more than anyone else, are responsible for the explosion of open research and improvement that has happened with things like llama.cpp that now allow you to run quite decent models locally on consumer h…

Why is Meta doing it though? This is an astronomical investment. What do they gain from it?

Zuck is pretty open about this in a recent earnings call:

https://twitter.com/soumithchintala/status/17531811200683049...

Re: Meta Llama 3

#524

Post by glenstein »

josh-sematic wrote: post #84
bbig wrote: post #4

They've got a console for it as well, https://www.meta.ai/ And announcing a lot of integration across the Meta product suite, https://about.fb.com/news/2024/04/meta-ai-assistant-built-wi... Neglected to include comparisons against GPT-4-Turbo or Claude Opus, so I guess it's far from being a frontier model. We'll see how it fares in the LLM Arena.

They also stated that they are still training larger variants that will be more competitive: > Our largest models are over 400B parameters and, while these models are still training, our team is excited about how they’re trending. Over the coming months, we’ll release multiple models with new capabilities including multimodality, the ability to converse in multiple languages, a much longer context window, and stronge…

Anyone have any informed guesstimations as to where we might expect a 400b parameter model for llama 3 to land benchmark wise and performance wise, relative to this current llama 3 and relative to GPT-4?

I understand that parameters mean different things for different models, and llama two had 70 b parameters, so I'm wondering if anyone can contribute some guesstimation as to what might be expected with the larger model that they are teasing?

Re: Meta Llama 3

#525

Post by refulgentis »

minimaxir wrote: post #72
refulgentis wrote: post #30

Earlier quoted context omitted.

Over yonder: https://x.com/alexalbert__/status/1780707227130863674 my $0.02: it makes me very uncomfortable that people misunderstand LLMs enough to even think this is possible

It is 100% possible for performance regressions to occur by changing the model pipeline and not the model itself. A system prompt is a part of said pipeline. Prompt engineering is surprisingly fragile.

Absolutely! That was covered in the tweet link. If you're suggesting they're lying*, I'm happy to extract it and check.

* I don't think you are! I've looked up to you a lot over last year on LLMs btw, just vagaries of online communication, can't tell if you're ignoring the tweet & introducing me to idea of system prompts, or you're suspicious it changed recently. (in which case, I would want to show off my ability to extract system prompt to senpai :)

Re: Meta Llama 3

#526

Post by glenstein »

medo-bear wrote: post #179
kleiba wrote: post #168

Earlier quoted context omitted.

What a silly, provocative comparison. China is a suppressive state that strives to control its citizens while the EU privacy protection laws are put in place to protect citizens. If you cannot access websites from "the free world" because of these laws, it means that the providers of said websites are threatening your freedom, not providing it.

[flagged]

>Nanny state is a nanny state.

In my opinion this is a thought stopping cliche that throws the concept of differences of scale out the window, which is a catastrophic choice to make when engaging in comparative assessments of policies in different countries. Again just my opinion here but I believe statements such as these should be understood as a form of anti-intellectualism.

Re: Meta Llama 3

#527

Post by zmmmmm »

sa-code wrote: post #122
flemhans wrote: post #99

Earlier quoted context omitted.

Probably the EU laws are getting too draconian. I'm starting to see it a lot.

EU actually has the opposite of draconian privacy laws. It's more that meta doesn't have a business model if they don't intrude on your privacy

They just said laws, not privacy - the EU has introduced the "world's first comprehensive AI law". Even if it doesn't stop release of these models, it might be enough that the lawyers need extra time to review and sign off that it can be used without Meta getting one of those "7% of worldwide revenue" type fines the EU is fond of.

[0] https://www.europarl.europa.eu/topics/en/article/20230601STO...

Re: Meta Llama 3

#528

Post by nilsherzig »

Manabu-eo wrote: post #463
buildbot wrote: post #87

Quick thoughts - Major arch changes are not that major, mostly GQA and tokenizer improvements. Tokenizer improvement is a under-explored domain IMO. 15T tokens is a ton! 400B model performance looks great, can’t wait for that to be released. Might be time to invest in a Mac studio! OpenAI probably needs to release GPT-5 soon to convince people they are still staying ahead.

> Might be time to invest in a Mac studio! The highest end Mac Studio with 196GB of ram won't even be enough to run a Q4 quant of the 400B+ (don't forget the +) model. At this point, one have to consider an Epyc for CPU inference or costlier gpu solutions like the "popular" 8xA100 80GB... An if it's a dense model like the other llamas, it will be pretty slow..

It's a dense one, zuck confirms this a couple minutes into the interview posted in this thread

Re: Meta Llama 3

#529

Post by michaelt »

brigadier132 wrote: post #453
tmalsburg2 wrote: post #324

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

Unsuccessful ideas can live on for a long time in a large corporation.

Nobody wants to tell the boss his pet project sucks - or to get their buddies laid off. And with Facebook's $100 billion in revenue, nobody's going to notice the cost of a few thousand engineers.

Re: Meta Llama 3

#530

Post by tinybear1 »

eru wrote: post #323
modeless wrote: post #89

I was curious how the numbers compare to GPT-4 in the paid ChatGPT Plus, since they don't compare directly themselves. Llama 3 8B Llama 3 70B GPT-4 MMLU 68.4 82.0 86.5 GPQA 34.2 39.5 49.1 MATH 30.0 50.4 72.2 HumanEval 62.2 81.7 87.6 DROP 58.4 79.7 85.4 Note that the free version of ChatGPT that most people use is based on GPT-3.5 which is much worse than GPT-4. I haven't found comprehensive eval numbers for the lates…

Via Microsoft Copilot (and perhaps Bing?) you can get access to GPT-4 for free.

* With targeted advertising
Post reply on HN