NinaCalvi/prefcollection-sample-skyworm-rm

该数据集包含多个字段，主要用于记录和比较不同文本生成模型的输出。字段包括prompt（提示）、chosen（选择的响应）、rejected（拒绝的响应）、truth_result（真实结果）、orig_instruction（原始指令）、orig_response_A（原始响应A）、orig_response_B（原始响应B）、messages（消息列表，包含内容和角色）、score_A（响应A的评分）、score_B（响应B的评分）。数据集包含3000个训练样本，总大小为35399703字节。

This dataset includes multiple fields primarily used for recording and comparing outputs from different text generation models. Fields include prompt, chosen (selected response), rejected (rejected response), truth_result (true result), orig_instruction (original instruction), orig_response_A (original response A), orig_response_B (original response B), messages (list of messages containing content and role), score_A (score for response A), and score_B (score for response B). The dataset contains 3000 training samples with a total size of 35399703 bytes.