mlfoundations-dev/instruction_filtering_scale_up_code_base_embedding_filter_mean_per_domain

该数据集包含了多个特征字段，如指令种子、来源、嵌入向量、正面评分平均值、负面评分平均值、差异分数、推理过程、deepseek解决方案、原始行索引和会话信息（包括发送者和信息内容）。数据集分为训练集，其大小为840,806,493字节，共有16,000个示例。数据集的总下载大小为403,252,133字节，总数据大小为840,806,493字节。

The dataset includes multiple feature fields such as instruction seed, source, embedding vectors, average positive score, average negative score, difference score, reasoning process, deepseek solution, original row index, and conversation information (including sender and message content). The dataset is split into a training set, which is 840,806,493 bytes in size and contains 16,000 examples. The total download size of the dataset is 403,252,133 bytes, and the total data size is 840,806,493 bytes.