正體中文Readme.md

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

• 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

WARNING: This draft is preliminary and subject to further changes.

To run this application, ensure you have the following dependencies installed on your system:

• Node.js v20.11.1 & npm
• PHP 8.1.27 & php-fpm & Composer
• Python 3.9.5 & pip
• Nginx or Apache
• Redis 6.0.20
• CUDA
• Git

Please follow these steps to set up and run the project on both Windows and Linux:

We have a portable version for Windows, only tested with Windows 10 x64. If you'd like to try it out easily, it's a good choice!

Otherwise, you can install the system using the following guide after ensuring that you have installed all the dependency programs listed above.

1. Clone the repository:

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

2. Install dependencies:

   • For Linux:

     cp .env.dev .env
     cd executable/sh
     ./production_update.sh
     cd ../kernel
     pip install -r requirement.txt

   • For Windows:

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

3. Configure PHP and PHP-FPM:

   • Make sure PHP is installed and configured correctly.
   • Configure your web server (Nginx or Apache) to serve the project's files. Set multi-chat/public as the html root.
   • Example setting file: multi-chat/nginx_config_example, multi-chat/php.ini
   • Recommended settings:
     • Set PHP max upload filesize to at least 10MB for RAG.
     • Set timeout of reading to at least 120 seconds or more for long-running models.

4. Configure Redis:

   • Ensure Redis Server is installed and running.
   • Configure it in .env or use default settings.
   • Start Redis worker by running php artisan queue:work --timeout=0 under multi-chat/ to process requests to the agent, It's recommanded to have at least 5 workers running at the same time.

5. Run the application:

   • Start your web server and PHP-FPM.
   • Run the agent in kernel/main.py with your Python installation. It's recommended to copy the agent folder to another location before execution.

6. Access the application:

   • First you need to seed a account, go multi-chat/ and run php artisan db:seed --class=AdminSeeder --force to seed your first admin account.
   • Open your web browser and go to the application's URL.
   • Login with your admin account and start using Kuwa GenAI OS

7. Setting up Models:

   • By default, there are no models installed. Please read this README to set up some models.
   • After setting up, the models will not appear on the website. Administrators must set the corresponding access_code on the website to access the model.
   • Please note that the Agent must be started before setting up the model (you can check if 127.0.0.1:9000 is accessible to confirm).

You can download the latest release of Kuwa GenAI OS for Windows and Linux.

Discord - Kuwa AI Discord community server

Facebook - Kuwa AI Community

Facebook - Kuwa AI 臺灣社群

Google Group - kuwa-dev

Facebook - Kuwa AI

Google Group - kuwa-announce

We're a small team of two, passionate about our project. If you're interested in what we've built, we'd love your contributions to help make our open-source project even better and more robust. Please don't hesitate to reach out if you're willing to lend a hand!

This software uses the following packages and programs:

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

Many thanks to Taiwan NSTC TAIDE project and AI Academy for their early supports to this project.

MIT