I have a set of well-structured data sitting in CSV files. The goal is to turn each row (or a selected subset) into a clear, flowing narrative that a large language model can read, “judge,” and score for effectiveness. I have drafted an initial prompt and sample output, but the wording feels clunky. I need your help refining that statement and, more broadly, shaping a repeatable conversion process. Here’s what I have in mind: • Ingest the CSV, pull out the key fields, and weave them into a readable paragraph or short story depending on the context. • Optimise the narrative so the LLM can reliably evaluate it for clarity, engagement, and fidelity to the source data. • Suggest or write a lightweight script (Python with pandas or another tool you like) that automates the conversion so I can run fresh data through the same pipeline. • Provide guidance on prompt structure so the judging step remains consistent across batches. Final deliverables: 1. A refined core prompt/statement ready for LLM evaluation. 2. A working conversion template or script with brief setup instructions. 3. Short usage notes explaining how to expand or tweak the process when the CSV schema changes. If you have experience turning tabular data into compelling prose and prompting models such as GPT-4 for scoring tasks, I’d love to see how you would tackle this.