Prompting well
A prompt is everything the model sees before it answers. Clear prompts get better answers: say what you want and who it's for, give it the context and material to use, show an example or two of the output you want (few-shot prompting; zero-shot means no examples), and ask it to work through hard problems step by step. Settings matter too: a higher temperature makes answers more varied, a lower one more predictable. A prompt only steers this answer; it doesn't change the model's weights.
Why it matters: Prompting is the cheapest way to get better results, so it's the first thing to try before RAG or fine-tuning.
Builds on: Why models make things up
Learn it, then prove it
Sign in (free) to take this step's mastery check and save your journey.
Contact Us
Questions or feedback about anything? We'd love to hear from you.
You can reach us at: support@seriousmindware.com