Learn Before
Benefit of Dynamic Few-Shot Examples
A developer is using a few-shot prompting strategy to have a language model classify user feedback into 'Positive', 'Negative', or 'Neutral'. They have a pool of 15 high-quality examples. Instead of using the same 3 examples in every prompt, they decide to randomly sample a different set of 3 examples from their pool for each new piece of user feedback they classify. What is the primary benefit of this approach compared to consistently using the same three examples for every classification?
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
Improving a Ticket Classification System
A developer is building a system to classify customer reviews. They have a collection of 20 high-quality examples of correctly classified reviews. They observe that the model's performance on new, tricky reviews changes significantly depending on which 3-4 examples are included in the prompt. To make the system more reliable and less dependent on any single, fixed set of examples, which of the following strategies directly addresses this specific observation?
Benefit of Dynamic Few-Shot Examples