You own little LLM in your matrix chatroom.

This project is split in two parts: the client and the server.

The server simply downloads an LLM and starts a llama-cpp-python server (which mimics an openai server).

The client connects to the matrix server and queries the llama-cpp-python server to create matrix messages.

• https://github.com/abetlen/llama-cpp-python
• https://github.com/ggerganov/llama.cpp/