wandb/RAGTruth-processed

RAGTruth数据集旨在评估文本生成模型中的幻觉现象，特别是在检索增强生成（RAG）环境中。数据集包含模型输出、专家注释、幻觉标签、质量评估和模型元数据。每个样本包含查询/问题、上下文段落、模型输出、幻觉标签（明显冲突和/或无根据信息）、质量评估和模型元数据（名称、温度）。数据集的注释由专家评审员完成，识别了两种类型的幻觉：明显冲突和无根据信息。数据集采用MIT许可证发布。

The RAGTruth dataset is designed for evaluating hallucinations in text generation models, particularly in retrieval-augmented generation (RAG) contexts. It contains examples of model outputs along with expert annotations indicating whether the outputs contain hallucinations. Each example contains a query/question, context passages, model output, hallucination labels, quality assessment, and model metadata. The dataset is divided into train and test splits, with statistics provided for each, including the distribution of hallucination and quality labels. The annotations were created by expert reviewers who identified two types of hallucinations: evident conflict and baseless information. The dataset is released under the MIT License.