What is the technique of providing examples inside a prompt often referred to as?	One-shot or multi-shot prompting.
How is "one-shot prompting" defined?	Providing a single example within the prompt.
How is "multi-shot prompting" defined?	Providing multiple examples within the prompt.
What structural method is recommended for wrapping input and output examples in a prompt?	Using XML tags.
When should multi-shot prompting be used to improve model performance?	To handle corner cases, such as sarcasm.
Besides handling corner cases, what is another key application of providing examples?	Helping the model understand complex output formats (e.g., generating complex JSON).
What is an optional but highly beneficial step when providing an ideal output example?	Including the reasoning or explanation for why that output is considered ideal.
In what context is providing examples particularly effective for prompt engineering?	When performing prompt evals.