Start Running LLaMA 3.1 405B In 3 Minutes With Ollama

preview_player
Показать описание
Meta just released an upgrade from LLaMA 3 to LLaMA 3.1 with a whopping new 405B parameter model variant! Thank Zuck for GPT-4o level intelligence that is open sourced. In this video, I'll show you how to deploy a multi-GPU cluster and how to set it up with only a few terminal commands. I hope this video helps and I appreciate you for watching!

► Allyson AI - Your AI Executive Assistant:

► FREE Guide For This Video:

►Tools Featured in This Video:
Ollama
Vast

► TIMESTAMPS:
0:00 - Intro to LLaMA 3.1 405B
0:23 - Start a Multi-GPU Cluster on Vast AI
1:34 - Install Ollama CLI on Linux
1:56 - How To Install LLaMA 3.1 405B
2:04 - LLaMA 3.1 405B Demo on Ollama
2:54 - Testing LLaMA 3.1 Jailbreak Prompt

► All My Links:

► VIDEOS YOU DON'T WANT TO MISS:
Рекомендации по теме
Комментарии
Автор

Your AI Executive Assistant - Hire Today For Only $49/mo

isaiahbjork
Автор

If this is what we get with 8 Gpu, then I am sticking to 70b for now.😢

abhijitramin
Автор

Thanks for the concise explaination. Simple and clear! (Comment for the algorithm).

TacticalCastle
Автор

I got a question, when I look in ollama the model requires aprox 335gb storage room but when I try to download it directly from meta the say it needs 750gb, why do you think is this difference?

king-zk
Автор

That's slow and not quite usable, what vast cluster should be faster and cost effective ?

carlosap
Автор

You said you would have the "jail break" prompt below.. I don't see it?

TacticalCastle
Автор

That's so slow, seems unusable in this way!

xcviij
Автор

Why is so Can you do something to spped up this??,

allo.allo.