Definition

Risk Function in Minimum Bayes Risk Decoding

In the context of minimum Bayes risk, a risk function, denoted as R(y,yr)R(\mathbf{y},\mathbf{y}_r), is defined for every pair of candidate outputs. This function quantifies the penalty or cost associated with replacing a chosen output y\mathbf{y} with an alternative reference output yr\mathbf{y}_r.

0

1

Updated 2026-04-30

Contributors are:

Who are from:

Tags

Ch.3 Prompting - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences