mlfoundations-dev/scale_up_science_50K_w_physics

该数据集包含多个字段，如文本内容(text)、类别(category)、URL(url)、标题(title)、问题(problem)、推理过程(reasoning)、解决方案(deepseek_solution)、推理轨迹(final_reasoning_trace)、对话(conversations)、学科(discipline)、专家(expert)、主题数(num_topics)、子主题数(num_subtopics)、问题数(num_questions)、主题(topic)、子主题(subtopic)、分数(score)和年份(year)。此外，数据集分为训练集(train)，包含50000个示例，大小为1.45GB。数据集的总下载大小为687MB。

The dataset includes multiple fields such as text content (text), category (category), URL (url), title (title), problem (problem), reasoning process (reasoning), solution (deepseek_solution), reasoning trace (final_reasoning_trace), conversation (conversations), discipline (discipline), expert (expert), number of topics (num_topics), number of subtopics (num_subtopics), number of questions (num_questions), topic (topic), subtopic (subtopic), score (score), and year (year). Additionally, the dataset is split into a training set (train) containing 50,000 examples, with a size of 1.45GB. The total download size of the dataset is 687MB.