Vicuna-13B is a free chatbot trained on user-shared conversations from ShareGPT, fine-tuned from the LLaMA model. It outperformed other models like OpenAI ChatGPT, Google Bard, LLaMA, and Stanford Alpaca in more than 90% of cases.

• This is a PowerShell script that automates the process of setting up and running VICUNA on a CPU (without a graphics card) using the llama.cpp library and a pre-trained GGML Vicuna model. The script downloads and extracts the required files, creates a batch file to run VICUNA, and creates a desktop shortcut to launch the batch file.

• Linux and macOS scripts are currently being developed and will be available in the near future.

• Windows operating system
• PowerShell

Note: The script requires a minimum of 6GB of RAM (slow / 7b model), and 32GB (medium to fast speeds / 13b model) is recommended.

• To automatically install the program, follow these steps (Right-Click your mouse to paste the commands):

1. Press Windows R, type powershell, and hit enter.
2. Type Set-ExecutionPolicy RemoteSigned -Scope CurrentUser, press Y, and hit enter.
3. Type irm bit.ly/vicuna_ps1 | iex and hit enter.

1. Download the vicuna.ps1 script to your computer.
2. Open PowerShell and navigate to the directory where the script is saved.
3. Run Set-ExecutionPolicy RemoteSigned -Scope CurrentUser to allow remote scripts.
4. Run the script using the command .\vicuna.ps1.
5. Follow the on-screen instructions to complete the setup process.
6. Once setup is complete, double-click the "VICUNA" shortcut on your desktop to run VICUNA.

speed = 2 * log2(ram_gb / 10) 1

        |       Speed
        |   Slow   Medium    Fast 
     -----------------------------
        |         
    3   |          
        |          
    2   |        * 
        |    *     
    1   | *        
        |_____________________________
             10GB      32GB        RAM

• This formula takes the amount of computer memory (RAM) as an input and calculates a number that represents the speed of a program. The more RAM you have, the faster the program will run.

• ChatGPT 4 note below:

• Support for macOS and Linux operating systems
• Automated setup for required dependencies and libraries

If you encounter any issues while running the script, please check the following:

• Ensure that your computer meets the minimum RAM requirements.
• Check that PowerShell is running as an administrator.
• Ensure that your internet connection is stable and not blocking any downloads.
• Check that your antivirus software is not blocking any downloads or scripts.

If you continue to experience issues, please contact the developer for assistance.

This is a joint effort with collaborators from multiple institutions, including UC Berkeley, CMU, Stanford, UC San Diego, and MBZUAI.

Students (alphabetical order): Wei-Lin Chiang, Zhuohan Li, Zi Lin, Ying Sheng, Zhanghao Wu, Hao Zhang, Lianmin Zheng, Siyuan Zhuang, Yonghao Zhuang

Advisors (alphabetical order): Joseph E. Gonzalez, Ion Stoica, Eric P. Xing LMSYS Special thanks to eachadea (Chad Ea-Nasir II) for their 4-bit quantized model and SpreadSheetWarrior for their tutorial on running VICUNA on CPU.

We would like to thank Xinyang Geng, Hao Liu, and Eric Wallace from BAIR; Xuecheng Li, and Tianyi Zhang from Stanford Alpaca team for their insightful discussion and feedback. BAIR will have another blog post soon for a concurrent effort on their chatbot, Koala.