OshanDeSilva/SiQuAD
收藏Hugging Face2026-04-22 更新2026-04-26 收录
下载链接：
https://hf-mirror.com/datasets/OshanDeSilva/SiQuAD
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是**SQuAD v1.0**数据集的僧伽罗语翻译版本，使用Google Translate API进行翻译。包含**16,000个问答对**，其中**13,000个用于训练**，**1,250个用于测试/开发**。数据集经过清洗和验证以确保翻译质量。主要用于僧伽罗语的抽取式问答模型训练和评估，最佳性能由XLM-R Large模型取得，F1得分为73.29。
This dataset is a translation of the **SQuAD v1.0** dataset into **Sinhala** using the Google Translate API. It consists of **16,000 question-answer pairs**, with **13,000 training pairs** and **1,250 test/dev pairs**. The dataset is cleaned and validated to ensure the quality of the translations. It can be used for training and evaluating extractive question answering models in Sinhala, with the best performance achieved by the XLM-R Large model (F1 score: 73.29).
提供机构：
OshanDeSilva
