ai4bharat/recon
收藏Hugging Face2024-11-27 更新2025-04-08 收录
下载链接：
https://hf-mirror.com/datasets/ai4bharat/recon
下载链接
链接失效反馈官方服务：
资源简介：
RECON测试集是一个精心策划的、人工注释的多语言评估基准，作为跨语言自动评估（CIA）套件的一部分推出。它旨在评估多语言大型语言模型（LLMs）的性能，并为评估器LLMs提供与人类判断对齐的基准。该数据集包含六种语言的目标语言提示和响应，以及英文参考答案和评估量表。
The RECON Test Set is a meticulously curated, human-annotated multilingual evaluation benchmark introduced as part of the Cross Lingual Auto Evaluation (CIA) Suite. It is designed to assess the performance of multilingual large language models (LLMs) and meta-evaluate evaluator LLMs. The dataset provides a comprehensive evaluation framework with prompts, responses, and scoring rubrics spanning six languages—Bengali, German, French, Hindi, Telugu, and Urdu—while maintaining reference answers and evaluation criteria in English.
提供机构：
ai4bharat
