mlfoundations-dev/math_stratos_scale_judged_and_annotated_with_difficulty

该数据集包含了一系列的问题以及对应的推理过程、解决方案等信息。每个问题都有其标准答案（ground_truth_solution）和用户提供的解决方案（deepseek_solution），并标注了用户答案的正确性（correct）。此外，数据集还包含了推理过程的评判（judge_reasoning）、问题的难度（difficulty）及难度的解释（difficulty_reasoning）。数据集分为训练集，共有140075个示例，文件大小为3,794,318,115字节。

The dataset consists of a series of problems along with corresponding reasoning processes, solutions, etc. Each problem has its standard answer (ground_truth_solution) and a solution provided by the user (deepseek_solution), along with a label indicating the correctness of the users answer (correct). In addition, the dataset includes judgments of the reasoning process (judge_reasoning), the difficulty of the problems (difficulty), and explanations for the difficulty levels (difficulty_reasoning). The dataset is split into a training set, which contains 140,075 examples and has a file size of 3,794,318,115 bytes.