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

The Llama 4 herd

ai.meta.com

Post reply on HN
Print view View original

381–390 of 695 posts

Re: The Llama 4 herd

#381

Post by AIPedant »

sshh12 wrote: post #129
ksec wrote: post #54

Interesting this is released literally one hour after another discussions suggesting Meta ( https://news.ycombinator.com/item?id=43562768 ) >at this point it does not matter what you believe about LLMs: in general, to trust LeCun words is not a good idea. Add to this that LeCun is directing an AI lab that as the same point has the following huge issues: 1. Weakest ever LLM among the big labs with similar resources (a…

Not that I agree with all the linked points but it is weird to me that LeCun consistently states LLMs are not the right path yet LLMs are still the main flagship model they are shipping. Although maybe he's using an odd definition for what counts as a LLM. https://www.threads.net/@yannlecun/post/DD0ac1_v7Ij?hl=en

[dead]

Re: The Llama 4 herd

#382

Post by steenandersson »

jhugg wrote: post #331
clueless wrote: post #128

Earlier quoted context omitted.

> Knowledge cutoff: August 2024. Could this mean training time is generally around 6 month, with 2 month of Q/A?

I wish my knowledge cutoff was August 2024.

This made me LOL louder than I have for a long time! Agree.

Re: The Llama 4 herd

#383

Post by popinman322 »

reissbaker wrote: post #346
kristopolous wrote: post #303

Earlier quoted context omitted.

17B puts it beyond the reach of a 4090 ... anybody do 4 bit quant on it yet?

Oh, it'll never run on a 4090. 17B is the active parameter count, not the total param count (and "active" doesn't mean you can slice just those params out and put them on the GPU — which parameters are active constantly changes, even per-token. "Active" just means you get tokens faster than a dense model). It's 109B total parameters, so you'd need at least 54.5GB VRAM just for the weights alone. A Framework Desktop,…

You can swap experts in and out of VRAM, it just increases inference time substantially.

Depending on the routing function you can figure out all the active experts ahead of the forward pass for a single token and pipeline the expert loading.

Re: The Llama 4 herd

#384

Post by aimanbenbaha »

qwertox wrote: post #63
laborcontract wrote: post #29

General overview below, as the pages don't seem to be working well Llama 4 Models: - Both Llama 4 Scout and Llama 4 Maverick use a Mixture-of-Experts (MoE) design with 17B active parameters each. - They are natively multimodal: text + image input, text-only output. - Key achievements include industry-leading context lengths, strong coding/reasoning performance, and improved multilingual capabilities. - Knowledge cuto…

Llama 4 Scout, Maximum context length: 10M tokens. This is a nice development.

I don't think RAG will survive this time

Re: The Llama 4 herd

#385

Post by AIPedant »

amrrs wrote: post #267

The entire licensing is such a mess and Mark Zuckerberg still thinks Llama 4 is open source! > no commercial usage above 700M MAU > prefix "llama" in any redistribution eg: fine-tuning > mention "built with llama" > add license notice in all redistribution

I am still dismayed how quickly we gave up on including the pre-training data as a requirement for "open-source" LLMs.

As someone who thinks LLMs as akin to Lisp expert systems (but in natural language): is like including the C source code to your Lisp compiler, but claiming the Lisp applications are merely "data" and shouldn't be included.

Re: The Llama 4 herd

#386

Post by annjose »

georgeck wrote: post #379
simonw wrote: post #343

This thread so far (at 310 comments) summarized by Llama 4 Maverick: hn-summary.sh 43595585 -m openrouter/meta-llama/llama-4-maverick -o max_tokens 20000 Output: https://gist.github.com/simonw/016ea0fd83fc499f046a94827f9b4... And with Scout I got complete junk output for some reason: hn-summary.sh 43595585 -m openrouter/meta-llama/llama-4-scout -o max_tokens 20000 Junk output here: https://gist.github.com/simonw/d01c…

I tried summarizing the thread so far (339 comments) with a custom system prompt [0] and a user-prompt that captures the structure (hierarchy and upvotes) of the thread [1]. This is the output that we got (based on the HN-Companion project) [2]: LLama 4 Scout - https://gist.github.com/annjose/9303af60a38acd5454732e915e33... Llama 4 Maverick - https://gist.github.com/annjose/4d8425ea3410adab2de4fe9a5785... Claude 3.7…

This is the script that assembles the structured comments and generates the summary - https://github.com/levelup-apps/hn-enhancer/blob/main/script...

You can run it as: node summarize-comments.js Example: node summarize-comments.js 43597782

And the summary will be put in the "output" folder.

You need to set the environment variable (in this case OPENROUTER_API_KEY because LLama4 is currently available at OpenRouter).

Re: The Llama 4 herd

#387

Post by jug »

Rover222 wrote: post #358
paxys wrote: post #330

Earlier quoted context omitted.

Should an "unbiased" model not create vikings of every color? Why offend any side?

It should be accurate. Adding in DEI to everything is a political bias. Truth is truth.

The problem here and with your comparison is that Gemini (the language model) wasn't creating black vikings because of political bias in the training, but due to how Google augmented the user prompts to force-include diversity. Behind the scenes, you were basically telling Gemini to always remember racial diversity even if you didn't in your prompt.

But if you were asking Gemini, vikings were white.

This was later rectified in an update once Google realized what mistake they had done, since it causes gross historical inaccuracies. But it wasn't rectified by doing anything to Gemini the language model. It did right all along.

Re: The Llama 4 herd

#389

Post by kristianp »

simonw wrote: post #343

This thread so far (at 310 comments) summarized by Llama 4 Maverick: hn-summary.sh 43595585 -m openrouter/meta-llama/llama-4-maverick -o max_tokens 20000 Output: https://gist.github.com/simonw/016ea0fd83fc499f046a94827f9b4... And with Scout I got complete junk output for some reason: hn-summary.sh 43595585 -m openrouter/meta-llama/llama-4-scout -o max_tokens 20000 Junk output here: https://gist.github.com/simonw/d01c…

Here's the link for model on openrouter: https://openrouter.ai/meta-llama/llama-4-maverick

Re: The Llama 4 herd

#390

Post by georgeck »

csdvrx wrote: post #370
simonw wrote: post #343

This thread so far (at 310 comments) summarized by Llama 4 Maverick: hn-summary.sh 43595585 -m openrouter/meta-llama/llama-4-maverick -o max_tokens 20000 Output: https://gist.github.com/simonw/016ea0fd83fc499f046a94827f9b4... And with Scout I got complete junk output for some reason: hn-summary.sh 43595585 -m openrouter/meta-llama/llama-4-scout -o max_tokens 20000 Junk output here: https://gist.github.com/simonw/d01c…

I have found the Gemini 2.5 Pro summary genuinely interesting: it adequately describes what I've read. Have you thought about automatizing hn-summaries for say what the 5 top posts are at 8 AM EST? That would be a simple product to test the market. If successful, it could be easily extended to a weekly newsletter summary.

This is a great idea! Exactly what I was also thinking and started working on a side-project. Currently the project can create summaries like this [1].

Since HN Homepage stories change throughtout the day, I thought it is better to create the Newsletter based on https://news.ycombinator.com/front

So, you are getting the news a day late, but it will capture the top stories for that day. The newsletter will have high-level summary for each post and a link to get the details for that story from a static site.

[1] - https://news.ycombinator.com/item?id=43597782

Post reply on HN