mlfoundations-dev/instruction_filtering_embedding_filter_math

该数据集包含了指令种子、来源、文本嵌入表示、最大正负分数、分数差、模型响应、原始行索引以及会话信息等字段。会话信息中包含会话的发送者和具体内容。数据集分为训练集，共有2500个示例，数据集大小为116,903,073字节。

The dataset includes fields such as instruction seed, source, text embedding representations, maximum positive and negative scores, score difference, model responses, original row index, and conversation information. The conversation information contains the sender and the content of the message. The dataset is split into a training set with a total of 2,500 examples, and the dataset size is 116,903,073 bytes.