Google

A practical technique in zero-shot learning involves repetitively adjusting prompts to guide a Large Language Model towards generating better responses. This iterative refinement process is conducted without providing the model with any explicit problem-solving steps or examples.

Iterative Prompt Adjustment in Zero-Shot Learning

A user is interacting with a large language model to generate a marketing slogan for a new brand of eco-friendly coffee. Below is a transcript of their first three attempts and the model's corresponding outputs. Analyze the user's strategy. In your analysis, identify the specific change the user made in each subsequent prompt and explain why this iterative process led to a more desirable final output.

Analyzing a Prompt Refinement Process

A user wants to generate a short, witty marketing slogan for a new brand of eco-friendly coffee beans. Arrange the following prompts in the most logical order of refinement, from the initial vague attempt to the final, most effective prompt.

A user wants a language model to generate a summary of a complex scientific article for a non-expert audience. The user's first attempt and the model's output are shown below:

**Initial Prompt:** "Summarize this article."

**Model's Output:** A highly technical summary filled with jargon, acronyms, and complex terminology directly from the article, making it difficult for a layperson to understand.

The user recognizes the output is not suitable for the intended audience. Which of the following

Learn Before

Related