zai-org/ChatGLM2-6B — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a Python project for running ChatGLM2 6B as a local bilingual chat assistant. I want someone to be able to install it, download or load the model from Hugging Face, then chat in Chinese or English through a simple web page, a command line demo, and an API that feels familiar for people who have used OpenAI style chat endpoints.

Please make it friendly to run on different machines, including options for lower memory use like int4 or int8 when possible. Keep the chat history working so it can handle follow up questions, and make the web demo clean and simple rather than fancy. Include clear setup steps, example prompts, and notes about hardware requirements and model limitations so users understand it may be wrong sometimes.

Also include a small evaluation or test script if that fits naturally, and organize the code so a developer can later add fine tuning or longer context support. Look up current docs online if you need to.

Want more depth? Deep Reverse