ThatsGroes/synthetic-from-retrieval-tasks-danish
收藏Hugging Face2025-01-24 更新2025-02-15 收录
下载链接:
https://hf-mirror.com/datasets/ThatsGroes/synthetic-from-retrieval-tasks-danish
下载链接
链接失效反馈官方服务:
资源简介:
该数据集包含100,000个样本,是为了预训练或后训练嵌入模型而设计,专门用于丹麦语的检索任务。数据集中的每个样本都是在特定的seed task基础上,使用gemma-2-27b-it模型生成的。样本包括提示(prompt)和模型输出(response),提示中包含内容(content)和角色(role)信息。
The dataset contains 100,000 samples and is designed for pre-training or post-training embedding models specifically for Danish retrieval tasks. Each sample in the dataset is generated based on a specific seed task using the gemma-2-27b-it model. The samples include prompts (prompt) and model outputs (response), where the prompts contain content (content) and role (role) information.
提供机构:
ThatsGroes



