TAUR-dev/linearized-solutions

该数据集包含了解题过程中的多个信息，如解题方案(solution)、问题(question)、思考轨迹(gemini_thinking_trajectory和deepseek_thinking_trajectory)、尝试(gemini_attempt和deepseek_attempt)、评分(gemini_grade和deepseek_grade)以及评分原因(gemini_grade_reason和deepseek_grade_reason)。此外，还包括了数据源类型(source_type)、元数据(metadata)、树结构信息(trees)和线性化解决方案(linearized_solution)。数据集分为训练集，可用于机器学习模型的训练。

The dataset includes multiple pieces of information about the problem-solving process, such as solution, question, thinking trajectories (gemini_thinking_trajectory and deepseek_thinking_trajectory), attempts (gemini_attempt and deepseek_attempt), grades (gemini_grade and deepseek_grade), and grade reasons (gemini_grade_reason and deepseek_grade_reason). It also includes source type (source_type), metadata, tree structure information (trees), and linearized solution (linearized_solution). The dataset is split into a training set, which can be used for training machine learning models.