[D] Training models when you have limited compute power

I've been wanting to take a code chatbot model like starchat or codellama and tune it to our codebase, problem is all I have at work is a Mac with 8gb of RAM. I talk with my boss today and can ask for some stuff if I want and can give good reason. What's the most efficient way to get the compute I need to train the model. Any other advice on how to go about doing this is greatly …

boss chatbot code codebase compute mac machinelearning power talk training training models work