Case Study

Calculating Risk for a Candidate Sentence

A language model is tasked with generating a one-sentence summary. To select the best option from several candidates, it uses a framework where a risk function, R(y,yr)R(y, y_r), quantifies the penalty for choosing a candidate summary, yy, when a reference summary, yry_r, is considered the ground truth. The specific risk function is defined as: R(y,yr)=1S(y,yr)R(y, y_r) = 1 - S(y, y_r), where S(y,yr)S(y, y_r) is a similarity score calculated as: (Number of unique words common to both summaries) / (Total number of unique words in the reference summary, yry_r). Based on the information provided below, calculate the risk value of choosing the candidate summary.

0

1

Updated 2025-09-29

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

Application in Bloom's Taxonomy

Cognitive Psychology

Psychology

Social Science

Empirical Science

Science