MLNTeam-Unical/MoralTextManipulation
收藏Hugging Face2025-07-10 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/MLNTeam-Unical/MoralTextManipulation
下载链接
链接失效反馈官方服务：
资源简介：
MoralTextManipulation 是一个大型数据集，旨在分析大型语言模型（LLM）在修订、重写或继续人类创作文本时如何影响道德表达。该数据集包括人类注释的文本和 LLM 操作的文本，涵盖各种任务和道德条件提示。它基于道德基础理论，并包括多个子集以供不同的研究问题使用。
MoralTextManipulation is a large-scale dataset designed to analyze how Large Language Models (LLMs) affect moral expressions when revising, rewriting, or continuing human-authored text. The dataset includes human-annotated texts and LLM-manipulated texts across various tasks and moral-conditioning prompts. It is based on the Moral Foundations Theory and includes multiple subsets for different research questions.
提供机构：
MLNTeam-Unical
