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

381–390 of 965 posts

Re: Meta Llama 3

#381

Post by zamadatix »

free_bip wrote: post #123
fnordlord wrote: post #95

Earlier quoted context omitted.

Do you mind my asking, if you're working on private codebases, how you go about using GPT/Claude as a code assistant? I'm just removing IP and pasting into their website's chat interface. I feel like there's got to be something better out there but I don't really know anyone else that's using AI code assistance at all.

Unless you have the privilege of being an enterprise customer with an SLA guaranteeing privacy, there's not much you can do other than using local models. I believe OpenAI says they don't train based on API requests but that's more of a "trust me bro" than any kind of guarantee.

Team and Enterprise come with the non-training guarantee, free and premium do not. Pretty much anyone can sign up for Team (I have, and I'm not a company) but you need to buy at least 2 seats for a total of $50/m. The rate limits are much better with that as well though.

Re: Meta Llama 3

#382

Post by lolinder »

cjbprime wrote: post #262
seydor wrote: post #108

Earlier quoted context omitted.

how much vram does the 8B model use?

In general you can swap B for GB (and use the q8 quantization), so 8GB VRAM can probably just about work.

If you want to not quantize at all, you need to double it for fp16—16GB.

Re: Meta Llama 3

#383

Post by freedomben »

mrtranscendence wrote: post #372
bigstrat2003 wrote: post #362

Earlier quoted context omitted.

Maybe you should consider that others may not share your views on Lex's lack of charisma or interesting thoughts.

I'll agree that "interesting thoughts" may be up to interpretation, but imma fight you on the charisma thing. I looked up "flat affect" in the dictionary and there were no words, only a full-page headshot of Lex Fridman.

Yeah, I'm a big fan of Lex because I think he is really good at building connections, staying intellectually curious, and helping peopl open up, but he is absolutely not big with charisma! I don't know if he normally talks so flat or not, but in the podcast I don't think he could be more flat if he tried. He's also not great at asking questions, at least not spontaneously. Seems really good at preparation though.

Re: Meta Llama 3

#384

Post by lordswork »

paxys wrote: post #229
bbig wrote: post #16

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

Very interesting part around 5 mins in where Zuck says that they bought a shit ton of H100 GPUs a few years ago to build the recommendation engine for Reels to compete with TikTok (2x what they needed at the time, just to be safe), and now they are accidentally one of the very few companies out there with enough GPU capacity to train LLMs at this scale.

TikTok (ByteDance) is now building an AGI team to train and advance LLMs (towards AGI), probably after realizing they are in a similar scenario.

Re: Meta Llama 3

#385

Post by bg24 »

janalsncm wrote: post #145
bg24 wrote: post #57

Last week, I made the explicit decision to circle through paid subscriptions of various providers every few months, instead of sticking to just ChatGPT. Glad to see Meta.ai on the list, in addition to Claude, Gemini, Mistral and Perplexity.

Any thoughts on which one you prefer? I have the ChatGPT subscription but curious about alternatives.

I have been using Claude and ChatGPT in parallel for 2 months now. Claude is slow to respond, but I love it. So I will stay on Claude for next 3-6 months for now. I find it unnecessary to pay $40 for both, but not use to the fullest.

So my plan is to use one chat subscription at any time, and then use APIs for the rest. Right now I have chatbox.app on my laptop, so have the freedom to play around w/ the rest as well as needed.

Re: Meta Llama 3

#386

Post by grepexdev »

elevatedastalt wrote: post #369
szundi wrote: post #353

Earlier quoted context omitted.

It is just the reverse: he is successful because he is like that and lots of founder ceos are jellies in comparison

I dunno. I find a conviction in passion in founder CEOs that is missing in folks who replace them. Compare Larry & Sergey with Pichai, or Gates with Balmer.

Don't forget Gavin Belson and Action Jack Barker

Re: Meta Llama 3

#387

Post by projectileboy »

paxys wrote: post #240
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…

You can see from Zuck's interviews that he is still an engineer at heart. Every other big tech company has lost that kind of leadership.

Anyone who made it through CS 121 is an engineer for life.

Re: Meta Llama 3

#388

Post by sunaookami »

blackeyeblitzar wrote: post #199
verdverm wrote: post #88

Initial observations from the Meta Chat UI... 1. fast 2. less censored than other mainstream models 3. has current data, cites sources I asked about Trump's trial and it was happy to answer. It has info that is hours old --- Five jurors have been selected so far for the hush money case against former President Donald Trump ¹. Seven jurors were originally selected, but two were dismissed, one for concerns about her im…

I recall there was a website tracking the ideological bias of LLMs, but I can’t find it now. But it was showing where all the LLMs rank on a political graph with four quadrants. I think we need something like that, ranking these LLMs on aspects like censorship. Example: https://www.technologyreview.com/2023/08/07/1077324/ai-langu... But I think some other site was doing this ‘live’ and adding more models as they appe…

Do you mean this? https://trackingai.org/

Re: Meta Llama 3

#389

Post by freehorse »

observationist wrote: post #67

https://github.com/meta-llama/llama3/blob/main/LICENSE Llama is not open source. It's corporate freeware with some generous allowances. Open source licenses are a well defined thing. Meta marketing saying otherwise doesn't mean they get to usurp the meaning of a well understood and commonly used understanding of the term "open source." https://opensource.org/license Nothing about Meta's license is open source. It's a…

What are the practical use cases where the license prohibits people from using llama models? There are plenty of startups and companies that already build their business on llamas (eg phind.com). I do not see the issues that you assume exist.

If you get that successful that you cannot use it anymore (have 10% of earth's population as clients) probably you can train your own models already.

Post reply on HN