Google

The empirical nature of prompt design necessitates an iterative, trial-and-error process to find an effective prompt. This involves creating an initial prompt, evaluating the model's output, and making successive adjustments until the responses align with the desired outcome.

Iterative Refinement of Prompts

A developer is working with a language model and is not getting the desired output. Arrange the following actions into the correct logical sequence for systematically improving the prompt.

A user is attempting to generate marketing ideas. After their first attempt resulted in a generic, unhelpful paragraph, they modified their input and received a much more useful, structured response. Analyze the specific changes made between 'Prompt 1' and 'Prompt 2' and explain why these modifications likely led to the improved output.

Analyzing Prompt Refinement

A user wants a language model to generate a short, professional email declining a meeting invitation. The user inputs the prompt: 'write an email'. The model produces a long, informal story about a person who dislikes meetings. Given this outcome, what is the most effective next step for the user to take to achieve their goal?

Learn Before

Related