AmanPriyanshu/stratified-kmeans-diverse-reasoning-100K-1M
收藏Hugging Face2025-10-04 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/AmanPriyanshu/stratified-kmeans-diverse-reasoning-100K-1M
下载链接
链接失效反馈官方服务：
资源简介：
这是一个从NVIDIA的Llama-Nemotron后训练数据集中精心平衡的子集，包含了数学、代码、科学、指令遵循、聊天和安全任务的多尺度多样子集。数据集使用了基于嵌入的k-means聚类方法来确保每个类别的多样性，并使用了平方根重平衡采样方法来调整不同类别的样本数量。
This is a carefully balanced subset of NVIDIAs Llama-Nemotron Post-Training Dataset, featuring square-root rebalanced sampling across math, code, science, instruction-following, chat, and safety tasks at multiple scales. The dataset uses embedding-based k-means clustering to ensure diversity within each category, and square-root rebalancing to adjust the sample count per category.
提供机构：
AmanPriyanshu
