adambuttrick/arxiv-author-affiliations-latex-extract-inference

该数据集包含与学术论文相关的信息，包括arXiv ID、DOI、文件名、标题（含文本和语言）、作者（包括标准作者和预测作者，以及所属机构）以及各种与预测相关的字段，如原始预测、思考预测、解析状态、输入/输出特性和处理指标。此外，还包含与模型运行相关的配置，如模型参数、性能指标和硬件信息。数据集分为四种配置，每种配置都有一个train分割，规模从非常小（5个示例）到较大（987个示例）不等。

The dataset contains information related to academic papers, including arXiv IDs, DOIs, filenames, titles (with text and language), authors (both gold standard and predicted, with affiliations), and various prediction-related fields such as raw predictions, thinking predictions, parsing status, input/output characteristics, and processing metrics. There are also configurations related to model runs, including model parameters, performance metrics, and hardware information. The dataset is split into four configurations, each with a train split, varying in size from very small (5 examples) to larger (987 examples).