banner
Darling

Darling

github
email

Local deployment of Llama3 - 8B/70B large models! The easiest way: support CPU/GPU running | 3 solutions

image

Currently in the field of open-source large models, Llama3 is undoubtedly the strongest! This time Meta not only released two powerful large models, 8B and 70B, for free, but also the 400B model is about to be released, which can compete with GPT-4! Today we will introduce 3 local deployment methods, simple and easy to understand, very suitable for beginners!

1. GPT4All: Suitable for low configuration users, can run on CPU/GPU Download here
image

2. LMstudio, supports downloading multiple models such as 8B/70B, more model choices! Download here
image

Reminder: If you are not overseas and cannot download the model, Llama 3 large models will be uploaded to a cloud drive later [Download link] to be updated later...

3. Ollama, supports multiple platforms! Can run on Windows/Mac/Linux Download here
image

And open the WSL feature on Windows

Install Docker Desktop: Go to download

Then install the webUI locally

  • Run on CPU:

docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main

  • Supports GPU running:
    docker run -d -p 3000:8080 --gpus=all -v ollama:/root/.ollama -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:ollama
Loading...
Ownership of this post data is guaranteed by blockchain and smart contracts to the creator alone.