RLHF-And-Friends/helpsteer3-multilingual

该数据集包含了多种语言的配置，包括中文、荷兰语、法语、德语、印度尼西亚语、意大利语、日语、韩语、波兰语、葡萄牙语、俄语、西班牙语和越南语。每个配置都包含了领域、语言、上下文（包括角色和内容）、响应1、响应2、整体偏好、个人偏好（包括分数、推理、反馈1、反馈2）和最佳响应。数据集还包括训练和验证分割，每个分割都有特定的示例数量和字节数，以及总的下载和数据集大小。

The dataset includes configurations for multiple languages such as Chinese, Dutch, French, German, Indonesian, Italian, Japanese, Korean, Polish, Portuguese, Russian, Spanish, and Vietnamese. Each configuration includes features such as domain, language, context (with role and content), response1, response2, overall_preference, individual_preference (with score, reasoning, feedback1, feedback2), and best_response. The dataset also includes training and validation splits with specific number of examples and byte sizes, as well as total download and dataset sizes.