로그인
통찰 - Discrimination Measurement in LLMs
Measuring Social Discrimination in Large Language Models: Prejudice and Caprice Framework
The author presents the Prejudice-Caprice Framework to comprehensively measure discrimination in Large Language Models by considering persistent prejudice and preference variation across diverse contexts.
Prejudice and Caprice: A Statistical Framework for Measuring Social Discrimination in Large Language Models
LLMs exhibit biases that can be quantified using the Prejudice-Caprice Framework, providing insights into discrimination risks.
Prejudice and Caprice: A Statistical Framework for Measuring Social Discrimination in Large Language Models
Prejudice-Caprice Framework (PCF) comprehensively measures discrimination in LLMs by considering persistent prejudice and preference variation across diverse contexts.
