Week two
Patterns:
- Solve a problem
- give me five ways to solve any problem that I give you and compare and contrast them. Ask me for the first problem.
Week Three
Five Components of a Prompt
- instructions
- information
- pattern/examples
- output format
- trigger Review: One, in-context learning. Yes, instructions can be really powerful ways of telling the large language model what we want. But sometimes just providing examples can be the most powerful way that we can get it to refine its behavior and pull out that behavior that we’re really looking for, those nuances.
Second, I want you to really remember that these models are powerful, so general tools for machine learning and AI. And you can go into classical machine learning and AI problems with these things, things like classification, sentiment, analysis. So make sure and know that and be aware of it and tap into it.
And then when you need to teach it something new, go back again to in context learning. Third, output formatting. Output formatting is one of the secret tools in an advanced prompt engineers tool belt.