Data question for fine tuning a gpt 4 model

Hi,
I work with rare events dataset (think protest events or conflict). I have been trying to fine tune a gpt-4 model with a prompt to annotate data. How often can I use the same dataset for fine-tuning? I want to try different parameter configurations: change batch size, epochs, LR multiplier. Can I use the same data or should I use a different dataset every single time?