aeolian83/HuggingFaceH4_ultrachat_200k_filtered_10k_sampled
收藏Hugging Face2025-04-09 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/aeolian83/HuggingFaceH4_ultrachat_200k_filtered_10k_sampled
下载链接
链接失效反馈官方服务：
资源简介：
这是一个用于Merge-Up SLM训练的对话数据集，包含prompt、prompt_id、对话消息（包括内容和角色）、token数量以及数据源等信息。数据集最初包含31388个样本，经过英语过滤和基于token长度的比例采样后，用于训练的数据集样本量为100,000。
This is a dialogue dataset for Merge-Up SLM training, containing prompt, prompt_id, chat messages (including content and role), token count, and dataset source. The dataset originally included 31,388 samples and was filtered for English content and sampled proportionally based on token length to reach a training dataset of 100,000 samples.
提供机构：
aeolian83
