Large language models (LLM) like ChatGPT are having a huge impact. They are also just the beginning. Over the next year, companies big and small will begin to roll out domain/persona specialized LLM models. Indeed, this is already becoming a reality with new products like the finance-specialized BloombergGPT and Microsoft’s developer-focused Copilot. We will soon see AI personal trainers, health coaches, councilors, legal assistants, and many more. While some cases may require fine-tuned models on domain-specific data, the majority can be accomplished with simple prompt engineering. But how do you know when your prompt is good enough? How can we generate objective accuracy scores on subjective text?

This guide will cover:

• Theory
• Prompt engineering
• Prompt Testing
• Prompt Scoring
• Prompt Feedback