CausalLM/Retrieval-SFT-Chat
收藏Hugging Face2025-02-14 更新2025-02-15 收录
下载链接:
https://hf-mirror.com/datasets/CausalLM/Retrieval-SFT-Chat
下载链接
链接失效反馈官方服务:
资源简介:
基于检索的多轮对话SFT合成数据集,包含10万条新合成的文本条目,支持英语、中文、日语和德语等多种语言,用于多轮对话的检索基础模型训练。数据集由长上下文模型生成,涵盖了多样化的主题,包括一些小众话题。数据集在2025年2月编译,使用大约620亿个token。
A Retrieval-Based Multi-Turn Chat SFT Synthetic Dataset containing 100k newly synthesized text entries, supporting multiple languages such as English, Chinese, Japanese, and German, designed for training retrieval-based dialogue models. The dataset is generated by long-context models, covering diverse themes including niche topics. Compiled in February 2025, the dataset uses approximately 62 billion tokens.
提供机构:
CausalLM



