Learn Before
Evaluating Prompts for AI Training
Consider two types of questions you could give to an AI model during a training phase where human reviewers will rank its multiple generated answers:
- 'What is the capital of France?'
- 'What are some fun and educational activities to do in Paris?'
Explain why one of these questions is significantly less effective than the other for this specific training purpose.
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
Imagine you are part of a team training a new AI assistant. A key step in this process involves providing the AI with a single question, generating multiple different responses to it, and then having human reviewers rank these responses from best to worst. This helps the AI learn what constitutes a high-quality answer. Which of the following questions would be most effective for this specific training step?
Optimizing an AI for Creative Brainstorming
Evaluating Prompts for AI Training