SamsungResearch/TRUEBench
收藏Hugging Face2025-10-15 更新2025-10-18 收录
下载链接：
https://hf-mirror.com/datasets/SamsungResearch/TRUEBench
下载链接
链接失效反馈官方服务：
资源简介：
TRUEBench是一个由Samsung Research提出的评估大型语言模型（LLM）作为人类工作助手性能的基准数据集，包含2400多个现实和具有挑战性的样本。数据集具有多语言特性，覆盖12种语言，并包含多种任务类型，如内容生成、编辑、数据分析等。数据集设计用于评估隐含约束，并能反映多人对话的场景。
TRUEBench is a benchmark introduced by Samsung Research to evaluate the performance of large language models (LLMs) as human job assistants, consisting of over 2,400 realistic and challenging samples. The dataset is multilingual, covering 12 languages, and includes various task types such as content generation, editing, data analysis, etc. It is designed to assess implicit constraints and reflect multi-turn conversation scenarios.
提供机构：
SamsungResearch
