OpenBMB/MiniCPM ? reverse-engineered prompt

Build me a simple but polished local app for MiniCPM5 so I can try the model on my own machine without a lot of setup pain.

I want a chat style demo with a clean interface, a place to load the 1B or 2B model, and easy options for running a normal version or a smaller quantized one. It should feel fast on a laptop, support streaming responses, and make it obvious how to start chatting, stop generation, and reset the conversation. If it makes sense, include a few useful example prompts for coding, math, and long context questions.

Also add clear setup steps in the README for installing, running the demo, and using the fine tuning and quantization parts of the project at a high level. Keep it friendly for someone who is not deeply technical, and if you need current package or model details, look them up online first.