Llama Lab is a starter repo for playing with open source LLMs.
I created it to easily spin up a new dev environment in persistent Cloud GPUs for running custom fine-tuning and inference jobs. It comes with a CLI utility you can use to parametrize a custom fine-tuning job using the dataset and model of your choice.
You can use it to reproduce the Alpaca paper or train your own Llama-derived LLM.