teamcore/DPO_Pm3B_U0_beta0.25dr_dpoEurus_RM_7bg

该数据集包含多个字段，如来源、指令、模型、完整性的子字段（如注释、评价、评分理由等）、批评、自定义系统提示、细粒度分数、模型类型、总体分数、原则、响应等。还包括正确和错误答案、提示、选定和拒绝的响应以及与不同模型相关的分数。数据集分为默认部分，提供了数据集的字节数和示例数。配置部分包含配置名称和数据文件路径信息。

The dataset includes multiple fields such as source, instruction, models, sub-fields under completions (like annotations, ratings, rationale for ratings, etc.), critique, custom system prompt, fine-grained score, model type, overall score, principle, response, etc. It also includes correct and incorrect answers, prompts, chosen and rejected responses, and scores related to different models. The dataset is split into a default section, providing information on the number of bytes and examples. The configs section contains configuration names and data file path information.