Diagnosing LLM Prompting Failures
A developer is trying to get a large language model to solve multi-step physics problems. They provide the model with a prompt that includes one example problem and its final numerical answer. However, when given a new problem, the model consistently calculates the wrong answer. Based on this scenario, explain the primary limitation of the developer's example.
0
1
Tags
Ch.2 Generative Models - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Ch.3 Prompting - Foundations of Large Language Models
Analysis in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
Chain-of-Thought (COT) Prompting
Explicitly Prompting for a Reasoning Process to Prevent Errors
A user wants a language model to solve a multi-step math word problem. The user's prompt includes an example of a different, but structurally similar, word problem along with its final numerical answer. Despite this example, the model fails to solve the new problem correctly. Which statement best analyzes the most probable cause of the model's failure?
Analyzing a Failed Prompt for a Logic Puzzle
Diagnosing LLM Prompting Failures