TAUR-dev/stapled-cots-dataset__gpt-4o-mini-annotator

该数据集包含了解决方案(solution)、问题(question)、cot类型(cot_type)、来源类型(source_type)等字段的信息。它还包括了元数据(metadata)、两种不同方法的思考轨迹(gemini_thinking_trajectory和deepseek_thinking_trajectory)、尝试次数(gemini_attempt和deepseek_attempt)、评分(gemini_grade和deepseek_grade)及评分原因(gemini_grade_reason和deepseek_grade_reason)。此外，还有采样cot(sampled_cots)和采样cot答案(sampled_cot_answer)。数据集被划分为训练集，共有1000个示例，总大小为73301581字节。具体用途和背景未在README中描述。

The dataset includes information fields such as solution, question, cot_type, source_type, metadata, two different methods thinking trajectories (gemini_thinking_trajectory and deepseek_thinking_trajectory), attempt counts (gemini_attempt and deepseek_attempt), grades (gemini_grade and deepseek_grade), and grade reasons (gemini_grade_reason and deepseek_grade_reason). It also contains sampled cot (sampled_cots) and sampled cot answers (sampled_cot_answer). The dataset is split into a training set with 1000 examples, totaling 73301581 bytes in size. The specific purpose and context are not described in the README.