A useful way to understand Model Training & Fine-Tuning is to separate the goal, the input, the model or tool, and the way the output is evaluated.
One important area is prompt design. Define what a good result looks like before changing tools or settings, then test one variable at a time so...