You can install LLaVA LLM for free on your local host. It’s free, open-source, and includes a vision feature. Let us show you how.
I found LLaVA last week and decided to try it out. I tested it in different ways and was really impressed with it.
I saw many people asking, “How can I download LLaVa LLM, install and use on my local machine?”. It’s a good question because you can’t use it if you don’t download it.
Quick steps to follow:
- Visit their GitHub repository.
- A few versions are offered: LLaVA 7B 8bit and LLaVA 7B 16bit.
- The 8-bit version is compatible with the free Google Colab, while the 16-bit requires Google Colab Pro for additional RAM.
Using Google Colab for LLaVA
• Save a copy to your Drive (which is a common step).
• Change the runtime type to ‘T4 GPU‘.
• Run the code:
– Clone the “LLaVA” GitHub repository
– Use the Python subprocess module to run the LLaVA controller
• Keep an eye on RAM and GPU usage during installation.
• Start a Gradio web server:
– The model runs on a specific port
– A public IP address is provided for access After the installation process, you should see this:
That’s it! “You’ll find the localhost link there after installation” – Click it and enjoy it!
Credits: Haider.
Read related articles: