正體中文 README_TW.md

Kuwa GenAI OS

An open, free, secure, and privacy-focused Generative-AI Operating System.

Key Features • Architecture • Installation Guide • Community • Acknowledgements • License

Key Features

• Multi-lingual turnkey solution for GenAI development and deployment on Linux and Windows

• Concurrent multi-chat, quoting, full prompt-list import/export/share and more for users

• Flexible orchestration of prompts x RAGs x bots x models x hardware/GPUs

• Heterogeneous supports from virtual hosts, laptops, PCs, edge servers to cloud

• Open source, allowing developers to contribute and customize the system according to their needs

Architecture

Warning: This draft is a preliminary version and may be subject to further changes.

Dependencies

To run this application, please make sure the following packages are installed on your system:

• Node.js v20.11.1 & npm
• PHP 8.1 & php-fpm & Composer
• Python 3.10 & pip
• Nginx or Apache
• Redis 6.0.20
• CUDA
• Git

For Windows and Linux, please follow the steps below to set up and run:

Installation Guide

If you want to try running fast to see, we provide a Portable Windows version and a Docker version, which have been tested in Windows 10 x64 and Ubuntu 22.04LTS environments. You can try them out!

Alternatively, you can refer to the following steps to install the entire system on your host. Before proceeding, please make sure you have installed all the dependencies listed above.

1. Clone the project:

   git clone https://github.com/kuwaai/genai-os.git
   cd genai-os/src/multi-chat/

2. Install dependencies:

   • For Linux:

     cp .env.dev .env
     cd executable/sh
     ./production_update.sh
     cd ../kernel
     pip install -r requirement.txt
     sudo chown -R $(whoami):www-data /var/www/html

   • For Windows:

     copy .env.dev .env
     cd executable/bat
     ./production_update.bat
     cd ../kernel
     pip install -r requirement.txt

3. Setting up PHP & PHP-FPM:

   • Make sure PHP is installed and configured properly.
   • Configure your web server (Nginx or Apache) to set src/multi-chat/public as the website root directory.
   • Example config files: nginx_config_example, php.ini
   • Recommended settings:
     • Set max upload file size in PHP to at least 10MB, for RAG applications.
     • Set Timeout to at least 120 seconds or more to avoid slower RAGs or lengthy model outputs.

4. Setting up Redis:

   • Make sure you have a Redis server installed and running.
   • Relevant settings can be adjusted in .env.
   • Run php artisan queue:work --timeout=0 under src/multi-chat/ to start the Redis Worker, which handles user requests. It is recommended to run at least 5 Redis Workers at the same time.

5. Run the application:

   • Start your web server and PHP-FPM.
   • Run the Kernel src/kernel/main.py. It is recommended to copy the Kernel folder to another location before running it.

6. Connect to the application:

   • First you need to create an admin account, go to src/multi-chat/, and run php artisan db:seed --class=AdminSeeder --force to seed your first admin account.
   • Open your browser and access the URL of your deployed Nginx/Apache application.
   • Log in with your admin account, and start using Kuwa GenAI OS!

7. Deploy models:

   • No model is provided by default. Please read this README to deploy some models.
   • After the deployment is complete, the models will not appear on the website out of thin air. The administrator must set the corresponding access_code on the website to access the model.
   • Please note that the Kernel must be started before deploying the model (you can check if you can connect to 127.0.0.1:9000 to confirm)

Download

You can download the latest version of Kuwa GenAI OS, supporting Windows and Linux.

Community

Discord - Kuwa AI Discord community server

Facebook - Kuwa AI Community

Facebook - Kuwa AI Taiwan community

Google Group - kuwa-dev

Announcement

Facebook - Kuwa AI

Google Group - kuwa-announce

Support

Our team currently has only two people. If you are interested in this project we are developing together, you can help us develop it together, help us make this open source project better. If you are willing to help, please do not hesitate to contact us at any time!

Packages & Applications

The following packages and applications are used in this project:

• PHP & PHP-FPM
• Laravel 10
• Python 3
• Node.js
• Docker
• Redis
• Marked
• highlight.js
• Nvidia CUDA

Acknowledgements

Grateful to the MOST TAIDE project and the Taiwan AI Academy for their assistance in the early development of this project.

License

MIT