README.md

local-llm

Local LLM enables you to run Large Language Models (LLMs) and other AI models locally on your computer without requiring access to the internet.

Features

Beautiful UI

It Actually Works

Built-in Model Browser

Built-in Model Downloader (With Progress Bars!)

Built-in Downloaded Model Manager

Installation

For the smoothest installation, you will need to install a virtual environment manager like Mamba.

To install Mamba, please refer to the official documentation here:

https://mamba.readthedocs.io/en/latest/installation/mamba-installation.html

Tip: If you are using Windows, add the following path to your PATH environment variable: C:\Users\Username\miniforge3\condabin, replacing Username with your current user profile's username. If you do not know how to do this, please search up "windows add to path".

Tip: If you are using Windows, run mamba init. Make sure to let the changes through your antivirus.

Tip: To stop conda from invasively activating itself upon every startup of your terminal application, run conda config --set auto_activate_base false.

We will be using Mamba to install most of our Python dependencies because pip is an extremely unintelligent and poorly programmed package manager that takes a long time to install packages.

First, create an environment and activate it using mamba:

mamba create -n "local-llm"
# Run activate with both mamba and conda in case mamba does not work.
mamba activate local-llm
conda activate local-llm

Then, install Node if you do not have it installed:

# If you do not have Node installed on your system, install Node below
mamba install -y nodejs # Installs Node and npm

Then, install backend dependencies with Mamba:

mamba install -y transformers accelerate flask waitress pytorch -c pytorch -c nvidia

Then, install frontend dependencies with npm:

npm install

Usage

First, ensure that the virtual environment for the project is activated:

# Run activate with both mamba and conda in case mamba does not work.
mamba activate local-llm
conda activate local-llm

Then, run

node start.js

If you ever make changes to the Javascript code of local-llm, please run

node start.js --build

to force local-llm to recompile the frontend and intermediate server layer.

If you are new, please navigate to the welcome page to get setup. We hope you enjoy using Local LLM!