Evaluating a Fine-Tuning Strategy

A development team is using supervised fine-tuning on a pre-trained language model to create a specialized legal document summarizer. After the initial training run, they find the model's summaries are factually accurate but often miss key nuances and critical legal arguments. Which of the following next steps best demonstrates the iterative experimentation and engineering effort central to optimizing the fine-tuning process?

A machine learning team is tasked with fine-tuning a pre-trained language model for a specific customer service chatbot application. Arrange the following actions into a logical, iterative sequence that reflects an effective optimization process.