A Gandhi-inspired LLM utilizing LLaMA 8B, hosted on GaiaNet for advanced AI interactions.
This project involves the creation of a Gandhi-based large language model (LLM) using the LLaMA 3 8B model, tailored specifically to reflect the principles and teachings of Mahatma Gandhi. At the core of this LLM is a modified system prompt, adjusted within the `config.json` file, where the instructions provided to the model are focused entirely on Gandhi's ideology—particularly his philosophy of non-violence (ahimsa), truth (satya), and peace.
To ensure efficient environment, the model is hosted on GaiaNet, an advanced platform, through the Windows Subsystem for Linux (WSL). The GaiaNet library has been integrated to optimize performance, enabling interaction. The choice of Gandhi's principles in the prompt ensures that the model's responses align with ethical and peaceful perspectives, inspired by one of the world's foremost leaders of human rights and social justice.
The project not only demonstrates the flexibility of LLaMA 8B but also explores the potential for AI to embody and propagate humanistic philosophies.
I made significant progress by closely following the documentation and setting up my own node on GaiaNet. I experimented with different models to determine the most suitable one for my project, ultimately settling on the LLaMA 3 8B model due to its ability to effectively represent Gandhi’s principles. However, I encountered performance issues due to my PC’s limited hardware (8GB RAM), which couldn’t handle the model's computational demands. To overcome this, I had to record the results and demonstrate the functioning LLM on a friend’s Mac.
THE PITCH VIDEO IS THE ONE WITH THE AUDIO No funds