copenlu/transparent-context-usage
收藏Hugging Face2025-10-09 更新2025-10-18 收录
下载链接：
https://hf-mirror.com/datasets/copenlu/transparent-context-usage
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个用于评估语言模型上下文利用率的标准数据集，特别是用于评估高亮解释能力是否能有效地反映模型对上下文的使用行为。数据集由四个常用的数据源组成，分为四种上下文设置，以测试高亮解释能否准确反映模型从哪个文档和哪个确切跨度推导答案。
This dataset is a benchmark for evaluating context utilisation in language models, especially for assessing whether the highlight explanation capability effectively reflects the models context utilisation behavior. The dataset is curated from four commonly used sources and organized into four context settings to test whether the highlight explanations accurately reflect which document and which exact span the model derives the answer from.
提供机构：
copenlu
