A human evaluator has ranked four machine-generated responses to a prompt in order of preference, from best to worst, as follows: Response D ≻ Response B ≻ Response A ≻ Response C. To create a training dataset, this single ranked list is converted into a set of pairs, where the first element of each pair is preferred over the second. Which of the following pairs would be an invalid entry in the resulting dataset?
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
A human evaluator has ranked four machine-generated responses to a prompt in order of preference, from best to worst, as follows:
Response D ≻ Response B ≻ Response A ≻ Response C. To create a training dataset, this single ranked list is converted into a set of pairs, where the first element of each pair is preferred over the second. Which of the following pairs would be an invalid entry in the resulting dataset?Calculating Pairwise Preference Dataset Size
Generating a Pairwise Preference Dataset