mlfoundations-dev/mix_base_code_math_t_0_25

该数据集包含了指令种子(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)和对话(conversations)等字段。其中对话字段是一个列表，包含发送者(from)和消息内容(value)。数据集被划分为训练集(train)，包含大约10000个样本，总大小约为587MB。

The dataset includes fields such as instruction_seed, source, reasoning, deepseek_solution, __original_row_idx, final_reasoning_trace, and conversations. The conversations field is a list that includes the sender (from) and message content (value). The dataset is split into a training set (train) with about 10,000 samples and a total size of approximately 587MB.