mlfoundations-dev/cqadupstack_seed_science

该数据集包含了多个字段，如标题(title)，文本(text)，问题(problem)，推理(reasoning)，解决方案(deepseek_solution)，推理跟踪(final_reasoning_trace)等。它还包含了一个对话列表(conversations)，其中包含对话的来源(from)和内容(value)。数据集分为训练集(train)，共有5000个样本，文件大小为255223147字节。此外，提供了默认配置(default)，指定了训练数据的文件路径。

The dataset includes multiple fields such as title, text, problem, reasoning, solution (deepseek_solution), final reasoning trace (final_reasoning_trace), etc. It also contains a list of conversations, which includes the source (from) and content (value) of the conversation. The dataset is split into a training set (train) with a total of 5000 samples, and the file size is 255223147 bytes. In addition, a default configuration is provided, specifying the file path for the training data.