mlfoundations-dev/instruction_filtering_scale_up_math_base_random_filtering_16K

这是一个包含对话和推理过程的数据集，其中包括指令种子(instruction_seed)、来源(source)、推理(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)等字段。数据集分为训练集(train)，共有16000个示例，大小为599740556字节。

This dataset contains conversations and reasoning processes, including fields such as instruction_seed, source, reasoning, deepseek_solution, __original_row_idx, and final_reasoning_trace. The dataset is split into a training set (train) with a total of 16,000 examples and a size of 599,740,556 bytes.