mlfoundations-dev/instruction_filtering_scale_up_code_base_random_filtering_2K

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)、原始行索引(__original_row_idx)、最终推理轨迹(final_reasoning_trace)以及对话(conversations)。对话字段包含对话的发起者(from)和对话内容(value)。数据集分为训练集(train)，共有2000个示例，总大小约为105MB。数据集的配置信息中，默认配置(default)指定了训练数据文件的路径。

The dataset includes multiple fields such as instruction_seed, source, reasoning, deepseek_solution, __original_row_idx, final_reasoning_trace, and conversations. The conversations field contains the initiator (from) and content (value) of the conversation. The dataset is split into a training set (train) with a total of 2000 examples, totaling approximately 105MB in size. The configuration information for the dataset specifies the path to the training data files under the default configuration (default).