mlfoundations-dev/seed_code_multiple_samples_majority_consensus_gpt_picker_seed_code_scale_up_2K

该数据集包含问题、来源、领域等信息，并且每个样本都有一个或多个70B模型的响应。此外，还包括了原始行索引、多数正确索引、判断理由、多数响应、验证后的响应以及对话信息。数据集被划分为训练集，共有4301个示例。

The dataset includes information such as problem, source, domain, and each sample has one or more responses from the 70B model. In addition, it contains the original row index, majority correct index, judgment reasoning, majority response, verified response, and conversation information. The dataset is split into a training set with a total of 4301 examples.