mlfoundations-dev/instruction_filtering_scale_up_code_base_random_filtering_16K

这是一个包含对话和推理过程的数据集，具体字段包括指令种子(instruction_seed)、来源(source)、推理过程(reasoning)、DeepSeek解决方案(deepseek_solution)、原始行索引(__original_row_idx)、最终推理轨迹(final_reasoning_trace)以及会话(conversations)信息。会话信息包括发送者(from)和消息内容(value)。数据集分为训练集(train)，共有16000个示例，总大小为842210928字节。

This dataset contains dialogues and reasoning processes, with fields including instruction seed (instruction_seed), source (source), reasoning (reasoning), DeepSeek solution (deepseek_solution), original row index (__original_row_idx), final reasoning trace (final_reasoning_trace), and conversation (conversations) information. Conversation information includes the sender (from) and message content (value). The dataset is split into a training set (train) with a total of 16,000 examples and a total size of 842,210,928 bytes.