mlfoundations-dev/instruction_filtering_scale_up_code_base_embedding_filter_mean_per_domain_8K

该数据集包含了多个字段，如指令种子、数据源、嵌入表示、正负评分、推理过程、解决方案、原始行索引和对话信息等。这些字段的数据类型包括字符串、浮点数和序列浮点数。数据集被拆分为训练集，其中包含了8000个示例。整个数据集的大小为624,660,879.5字节。具体应用场景和详细内容未在README中描述。

The dataset includes multiple fields such as instruction seed, source, embedding representations, positive and negative scores, reasoning process, solutions, original row index, and conversation information. The data types of these fields include strings, floating-point numbers, and sequence of floats. The dataset is split into a training set, which contains 8000 examples. The total size of the dataset is 624,660,879.5 bytes. The specific application scenario and detailed content are not described in the README.