Learn Before
Evaluating a Constrained Prompt Search Space
A team is developing an automated system to find the best prompt for a language model that generates creative story plots. They define the set of all possible prompts the system can test using a strict template: 'Write a story about a [CHARACTER] who wants to [GOAL] but faces the obstacle of [OBSTACLE].' The system can only replace the bracketed placeholders with words from predefined lists, each containing 20 options (20 characters, 20 goals, 20 obstacles).
Critically evaluate this method of defining the boundaries for the prompt search. What is a major strength and a major weakness of this approach for the specific task of fostering creativity?
0
1
Tags
Ch.3 Prompting - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Evaluation in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
Example of Prompt Search Space Construction
A team is using an automated algorithm to discover the best prompt for a language model that summarizes complex scientific papers into a single paragraph for a general audience. To do this, they must first define the set of all possible prompts the algorithm can test. Which of the following strategies for defining this set is most likely to lead to the discovery of a highly effective prompt?
Evaluating a Prompt Optimization Strategy
Evaluating a Constrained Prompt Search Space