mlfoundations-dev/science_verification_scale_up_gpt_verification

该数据集包含多个字段，如问题（problem）、推理（reasoning）、解决方案（deepseek_solution）、原始行索引（__original_row_idx）、最终推理跟踪（final_reasoning_trace）、正确性（correct）、分类器推理（classifier_reasoning）、多数响应（_majority_responses）、验证的最终推理跟踪（verified_final_reasoning_trace）和对话（conversations）。数据集分为训练集（train），共有43607个示例。数据集的总大小为3365003830字节。

The dataset includes fields such as problem, reasoning, solution (deepseek_solution), original row index (__original_row_idx), final reasoning trace, correctness (correct), classifier reasoning, majority responses (_majority_responses), verified final reasoning trace, and conversations. The dataset is split into a training set (train) with a total of 43,607 examples. The total size of the dataset is 3,365,003,830 bytes.