mlfoundations-dev/instruction_filtering_scale_up_code_base_random_filtering_4K

该数据集包含多个字段，如指令种子(instruction_seed)、来源(source)、推理(reasoning)、解决方案(deepseek_solution)等。数据集还记录了原始行索引(__original_row_idx)和最终推理跟踪(final_reasoning_trace)。此外，数据集还有一个对话(conversations)列表，其中包含对话的发送者(from)和内容(value)。数据集分为训练集(train)，共有4000个示例，大小为210552732字节。

The dataset includes multiple fields such as instruction_seed, source, reasoning, deepseek_solution, etc. It also records the original row index (__original_row_idx) and the final reasoning trace (final_reasoning_trace). In addition, the dataset has a list of conversations, which includes the sender (from) and content (value) of the conversation. The dataset is split into a training set (train) with a total of 4000 examples and a size of 210552732 bytes.