Prompt Engineering
Prompt Engineering We'll learn about Prompting best practices for software development Two types of LLMs: Base LLM: Predicts next word, based on text training data Instruction tuned LLM: Where lot of research and practice has been going on. We fine tune on the instructions and LLM makes good attempts to follow those instructions. LLMs have been trained to be helpful, honest, harmless. Let's focus on the best practices for Instruction tuned LLMs. When you use an instruction tuned LLM, think of giving instructions to some other person. We'll see examples of how to be clear and specific which is important principle for prompting LLMs. 2nd principle is Giving the LLM time to think. To summarize, 1st principle: clear and specific instructions 2nd principle: give the model time to think.. Prompt Engineering Tactics: 1st Principle : Be clear and specific Note: clear doesn't necessarily mean short. So it could be long as well. Tactic1: Use delimiters to indicate distinct part...