TAUR-dev/lmfd_eval_r1_8d__steps_with_stats

该数据集包含了问题、解决方案、评估提示及其内容与角色、内部上下文、评估解决方案、原始评估提示、正确性判断和推理判断等字段。评估步骤详细记录了是否包含最终答案、内容、方程、是否完成、输出、步骤类型、子目标和验证子目标等信息。此外，还包含了步骤和跟踪的单词数差异、评估步骤的统计信息。数据集分为训练集，其大小为10,593,941字节，共有99个示例。

The dataset includes fields such as questions, solutions, evaluation prompts and their content and roles, internal context, evaluation solutions, raw evaluation prompts, correctness judgment, reasoning judgment, etc. The evaluation steps detail whether they contain the final answer, content, equation, whether it is finished, output, step type, subgoal, and verification subgoal, etc. Additionally, it includes the word count difference between steps and traces, as well as statistics for evaluation steps. The dataset is split into a training set, which is 10,593,941 bytes in size and contains 99 examples.