aeolian83/allenai_tulu_3_sft_mixture_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/allenai_tulu_3_sft_mixture_filtered_10k_sampled
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个用于Merge-Up SLM训练的采样数据集，包含3058个原始样本。通过过滤仅保留包含英语字母的样本，并根据样本的token长度进行比例采样，最终构建了一个包含100,000个样本的数据集。
This dataset is a sampled dataset for Merge-Up SLM training, containing 3058 original samples. By filtering to retain only samples with English alphabets and sampling proportionally based on the token length of samples, a dataset of 100,000 samples was constructed.
提供机构：
aeolian83
