Setup Tutorial and sample Run

Hello, I've spent a last few days playing with this repo and it's submodules, and I have definitely learned a lot but I'm still not at the point where I can run on my CUDA GPU machine the code.

My goal is to be able to run something like this demo:

https://llemma-demo.github.io/

"MATH Algebra Level 3" and put in print statements to show detailed debug info about how the probabilities are computed and the right tokens selected.

Anyone interested in making a step by step tutorial? I'm doing this for a client so I can pay.