deepseek-ai/DeepSeek-R1 ? reverse-engineered prompt

Hey, I want to play around with the new DeepSeek-R1 AI model. I saw it on GitHub and it looks really powerful—apparently it's great at reasoning, math, and even coding tasks.

Can you set up a local project for me where I can use it? I'd like a simple Python script or maybe a Jupyter notebook where I can input a text prompt and get a response from the model. The repository mentions a few different versions, including some smaller 'distilled' models based on Llama and Qwen that are still very capable. Let's start with one of those, maybe a 7B or 8B parameter one. You'll have to handle downloading it from Hugging Face and setting up all the necessary dependencies. The README mentions a 'Usage Recommendation' section, so please check that and follow any specific instructions from the maintainers.