tomaarsen/gooaq-reranker-blogpost-datasets

该数据集包含三种配置：hard-labeled-pair、pair和rerank。hard-labeled-pair配置包含问题、答案和标签特征，适用于训练有监督的模型。pair配置包含问题、答案特征，有训练集和评估集，适用于训练和评估模型。rerank配置包含问题、答案和多个负样本特征，只有评估集，适用于重排任务的模型评估。

The dataset consists of three configurations: hard-labeled-pair, pair, and rerank. The hard-labeled-pair configuration includes question, answer, and label features, suitable for training supervised models. The pair configuration includes question and answer features with both training and evaluation sets, suitable for training and evaluating models. The rerank configuration includes question, answer, and multiple negative sample features, with only an evaluation set, suitable for model evaluation in re-ranking tasks.