tum-nlp/cognitive-biases-in-llms
收藏Hugging Face2025-10-30 更新2025-11-15 收录
下载链接：
https://hf-mirror.com/datasets/tum-nlp/cognitive-biases-in-llms
下载链接
链接失效反馈官方服务：
资源简介：
用于评估大型语言模型中认知偏见的存在和强度的表格数据集。该数据集包含30种认知偏见，每种偏见有1000个测试案例，总共30000个测试案例，旨在探测大型语言模型在管理决策情景中的偏见表现。
A tabular dataset for measuring the presence and strength of cognitive biases in large language models (LLMs). The dataset contains 30 types of cognitive biases, with 1,000 test cases for each bias, totaling 30,000 test cases, designed to probe the bias behavior of LLMs in managerial decision-making scenarios.
提供机构：
tum-nlp
