textdetox/detoxification_pairwise_content_evaluation
收藏Hugging Face2025-07-22 更新2025-08-09 收录
下载链接：
https://hf-mirror.com/datasets/textdetox/detoxification_pairwise_content_evaluation
下载链接
链接失效反馈官方服务：
资源简介：
该数据集旨在用于微调大型语言模型，以评估解毒质量——即生成的文本是否与原始文本相似。数据集特别提供了文本对是否相似的答案：是或否。数据集来源于两个共享任务的注释数据集，包括RUSSE 2022和TextDetox CLEF 2024，并包含了多个团队的提交数据。
The dataset is aimed at fine-tuning large language models (LLM) to evaluate the quality of detoxification - whether the generated text is similar to the original one. Specifically, the dataset provides answers whether the pair of texts is similar: yes or no. The data source is from annotated datasets of two shared tasks, including RUSSE 2022 and TextDetox CLEF 2024, and includes submissions from several teams.
提供机构：
textdetox
