PatentDesc-355K

该数据集是一个大规模的数据集，包含了大约355,000个专利图表，以及从超过60,000份美国专利文件中提取的简短和详细的文本描述。此外，该数据集确保了训练集与验证集/测试集之间的独立性，并包含了详细的统计数据和质量评估指标。其规模达到了355,000个专利图表，任务是对专利图表生成描述。

This is a large-scale dataset containing approximately 355,000 patent figures, as well as both concise and detailed textual descriptions extracted from over 60,000 U.S. patent documents. Furthermore, this dataset ensures the independence between the training set and the validation/test sets, and includes comprehensive statistical data and quality assessment metrics. The core task of this dataset is to generate textual descriptions for patent figures, with a total of 355,000 patent figures included.