Meta lets Llama 3 LLM out to graze, claims it can give Google and Anthropic a kicking

Meta has unleashed its latest large language model (LLM) – named Llama 3 – and claims it will challenge much larger models from the likes of Google, Mistral, and Anthropic. Revealed in a lengthy announcement on Thursday, Llama 3 is available in versions ranging from eight billion to over 400 billion parameters. For reference, …

[Dostoevsky]

Oh, great...

Another "Plagiarized Information Synthesis System."

I'll let y'all figure out the acronym. :)

[AdmFubar]

the real issue is,who will "Whip the llama's ass"

[Korev]

Media players for the WinAmp!

[AdmFubar]

The real issue will be, who will "whip the llama's ass"?

[Management Order]

Watch out for the Llamatron! It will get you!

https://en.wikipedia.org/wiki/Llamatron

[Bebu]

Llama3 giving google and anthropic a good kicking... AI meets Clockwork Orange

The image that links to this article looks like one of a flock of demented, cyber converted sheep with which John Lumic was going to conquer the world.*

The fictional Lumic is arguably a lot saner than most of the loonies in charge of some of largest enterprises on the planet.

Another Aggravated Insanity Large Lying Machine is all we need as though we don't already have an excess of the natural sort.

* sheep are dangerous and, arguably worse, that llamas are just camels hiding in sheeps' clothing.

[Richard Tobin]

Running it with ollama

Somewhat surprisingly I can run the 70b model on my 32GB Mac. As you'd expect, it's very slow: output takes about a minute per token.

The 8b model runs at a reasonable speed.

[druck]

No more

Let's hope they find the 400bn parameter version is no better than the 70bn one before they burn the entire planet for electricity.