Learn Before
A developer is trying to get a language model to solve a specific type of multi-step logic puzzle. The model consistently fails when given just the puzzle. To improve performance, the developer constructs a prompt that first shows two different puzzles of the same type, each followed by a detailed, step-by-step explanation of how to arrive at the correct solution. Finally, the prompt presents the new puzzle for the model to solve. Why is this prompting strategy effective for this task?
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
Analysis in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
Example of a Few-Shot CoT Prompting Demonstration (Boris and Beck's Apples)
A developer is trying to get a language model to solve a specific type of multi-step logic puzzle. The model consistently fails when given just the puzzle. To improve performance, the developer constructs a prompt that first shows two different puzzles of the same type, each followed by a detailed, step-by-step explanation of how to arrive at the correct solution. Finally, the prompt presents the new puzzle for the model to solve. Why is this prompting strategy effective for this task?
Constructing a Reasoning Demonstration
Evaluating Prompting Strategies for Reasoning Tasks