mlfoundations-dev/a1_science_pubmed_science

该数据集包含多个字段，如发布ID（pubid）、指令种子（instruction_seed）、上下文信息（context）、长答案（long_answer）、最终决策（final_decision）等。上下文信息包括多个序列，如文本上下文（contexts）、标签（labels）、网格（meshes）、无推理预测（reasoning_free_pred）、需要推理的预测（reasoning_required_pred）等。数据集分为训练集，包含40000个示例，总大小为433,706,745字节。

The dataset includes multiple fields such as publication ID (pubid), instruction seed (instruction_seed), context information (context), long answers (long_answer), final decisions (final_decision), etc. The context information includes multiple sequences such as text contexts (contexts), labels (labels), meshes (meshes), prediction without reasoning (reasoning_free_pred), prediction requiring reasoning (reasoning_required_pred), etc. The dataset is split into a training set, containing 40,000 examples, with a total size of 433,706,745 bytes.