Justifying the Use of Multiple Demonstrations
A machine learning engineer is designing a prompt to extract specific financial data (e.g., revenue, net income) from unstructured company earnings reports. They observe that when they provide only one example of a report and its extracted data, the model often misses data points or extracts incorrect information from new reports. Explain the reasoning behind why providing three to five diverse examples, instead of just one, would likely lead to more accurate and reliable data extraction.
0
1
Tags
Ch.3 Prompting - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Analysis in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
A developer is prompting a large language model to convert customer reviews into a structured summary containing only the product's pros and cons. The initial prompt includes a single example of a review and its corresponding summary, but the model's outputs are inconsistent. Which of the following changes to the prompt is most likely to improve the model's reliability in generating the desired structured summaries?
Improving a Ticket Classification Prompt
Justifying the Use of Multiple Demonstrations
Example of a Few-Shot Prompt for Grammar Correction