ZixuanKe/flare_finqa_sup_sample_from_policy_v1.1_stepwise_dpo_chunk_6

该数据集用于训练模型，包含用户提示（prompt）、被拒绝的回答（rejected）、被接受的回答（chosen）以及选择该回答的解释（justification）。数据集仅包含一个训练集，共有1642个样本。

This dataset is used for training models and includes user prompts (prompt), rejected responses (rejected), chosen responses (chosen), and justifications for selecting those responses (justification). The dataset consists of a single training set with 1642 samples.