What is a key characteristic of prompt evaluation workflows regarding methodology?	There is no single standard methodology; workflows can be assembled in many different ways.
What is the first step in a typical prompt evaluation workflow?	Writing an initial prompt draft.
What is the purpose of the evaluation data set?	To contain a list of possible inputs (e.g., questions) that will be fed into the prompt.
After combining the prompt and the data set, what action is taken?	Each fully fleshed-out prompt is fed into the LLM (e.g., Claude) to generate a response.
What is the function of the "grader" in the evaluation workflow?	To take the original input/question and the LLM's response and assign a quality score (e.g., 1 to 10).
How is the objective performance metric of the prompt determined?	By averaging the scores generated by the grader across all test cases.
What is the goal of iterating through the evaluation process?	To modify the prompt based on the scores and repeat the pipeline to improve performance.