RLAIF/dpo_thinking_n_a_o_h_u_p_1e-6_0.02_1.7B_1.7B_with_gold_labels_kl_estimation
收藏数据链接：
官方服务：
资源简介：
该数据集包含了一系列的问题和答案，以及它们之间的相似度度量（KL散度）。每个问题都有多个字段，包括步骤编号、问题文本、参考答案、当前答案等。数据集被划分为训练集，可用于训练模型以理解和评估答案的质量。
The dataset consists of a series of questions and answers along with their similarity measurements (KL divergence). Each question has multiple fields including step number, question text, reference answer, current answer, etc. The dataset is split into a training set, which can be used to train models to understand and evaluate the quality of answers.
提供机构：
RLAIF
