mlfoundations-dev/instruction_filtering_scale_up_math_base_embedding_filter_mean_per_domain

该数据集包含了多个特征字段，如指令种子、数据来源、嵌入向量、正面评分平均值、负面评分平均值、评分差值、推理过程、DeepSeek解决方案、原始行索引和最终推理跟踪。此外，还包含了对话信息，每个对话有发送者和信息内容。数据集划分为训练集，共有16000个示例，总大小约为1.1GB。

The dataset includes multiple feature fields such as instruction seed, source, embeddings, mean positive score, mean negative score, difference score, reasoning, DeepSeek solution, original row index, and final reasoning trace. It also contains conversation information with each conversation having a sender and message content. The dataset is split into a training set with a total of 16,000 examples and is approximately 1.1GB in size.