Most takes on this are so bad. LLMs are trained on trillions of parameters to build a model. This model can work on your data, summarize and study your stuff. But you can’t just train a new model on your data, this just wouldn’t make a smart LLM.