spike: Use Flock.io to run an experiment to train an LLM on Masa data

[teslashibe]

Problem

We want to understand the value and utility of our data by using Flock.io and Cortex.t to train and test pre-trained LLMs in order to evaluate the efficacy of our data in a model. We do not know what data with what model has the highest value to our users.