meta-llama/llama ? reverse-engineered prompt

Hey, can you build me a simple Python project to run Meta's Llama 2 models locally on my machine? I want to be able to experiment with them myself.

It should support two main functions. The first is a basic text completion tool, where I can provide a starting prompt and it will continue writing from there. The second should be a chat completion mode, allowing me to have a back-and-forth conversation, like a simple chatbot.

I know I'll have to go to the Meta website and register to get the actual model weights and a special download link. Please include a script that uses that link to download the necessary files. After that, I just want simple example scripts to run either the text completion or a chat session, letting me point the code at whichever model checkpoint I've downloaded.