mlfoundations-dev/instruction_filtering_embedding_filter_mean_seed_data_code_w_openthoughts

这是一个包含对话和问题解决信息的数据集，其中包括系统信息、对话内容（包括发送者和消息内容）、问题、解决理由、DeepSeek解决方案、ID、真实解决方案、来源、正确性、判断理由、原始行索引、指令种子、嵌入向量、正面平均分数、负面平均分数和分数差等字段。数据集分为训练集，共有120668个示例。

This is a dataset containing conversation and problem-solving information, including fields such as system information, conversation content (including sender and message content), problem, reasoning, DeepSeek solution, ID, ground truth solution, source, correctness, judge reasoning, original row index, instruction seed, embeddings, mean positive score, mean negative score, and difference score. The dataset is split into a training set with a total of 120668 examples.