EleutherAI/bergson-asymmetric-style

该数据集包含多个特征字段，包括事实(fact)、重述(reworded)、答案(answer)、领域(field)、标识符(identifier)、问题(question)、模板(template)、风格(style)和预期匹配风格(expected_match_style)。数据集分为训练集(train)和多个评估集(eval, eval_majority_style, eval_original_style, eval_pirate_style)，训练集包含13,500个样本，每个评估集包含4,500个样本。数据集总大小为8,904,067字节，下载大小为2,188,163字节。

The dataset includes multiple feature fields such as fact, reworded, answer, field, identifier, question, template, style, and expected_match_style. It is divided into a training set (train) and several evaluation sets (eval, eval_majority_style, eval_original_style, eval_pirate_style). The training set contains 13,500 examples, and each evaluation set contains 4,500 examples. The total dataset size is 8,904,067 bytes, with a download size of 2,188,163 bytes.