How To Run Llama 3.1: 8B, 70B, 405B Models Locally (Guide)

preview_player
Показать описание

Meta has finally released Llama 3.1 models which feature: 8B, 70B and a massive 405B model which is able to compete with the state-of-the-art closed source models. In this video I will show you how to run and deploy Llama 3.1 models locally on your laptop, along with their requirements.

Рекомендации по теме
Комментарии
Автор

good, I will now try to run the 405B model on my 50, 000$ pc.

ecchiRhino
Автор

You can run the 405b model on a server with at least 256gb of ram and a vega 56 or 64 graphics card.
That specific card can access ram as if it were vram and on some platforms, bypass the CPU.
You can also buy optane pm for cheap. Either adding 4x4 optane modules via pcie then raid 0 and add the raid to swap, or as ddr4 dims to just add as a massive ram pool.
I'll be experimenting with Vega 20 next week

flpiutc
Автор

um 405b is like 300 gigs good luck with THAT lol

gu
Автор

Thanks a lot for the detailed explanation in the video! I have a question regarding Ollama. Is it possible to use Ollama and the models available on it in a production environment? I would love to hear your thoughts or any experiences you might have with it. Thank you!

Автор

is there a single source of Information which could give details hardware requirements for each of Llama 3.5 Models (i.e. GPU, RAM, Memory, Cache Memory etc.)

deepaksingh
Автор

Have u guys tried 405b version localy ? What are ur pc specs ?

ouso