few_rel

FewRel 旨在提供大规模的小样本关系抽取能力，它包含超过一百种关系以及数万个跨不同领域的标注实例。该仓库的数据集规模在1万到10万条样本之间，主要由英文文本和众包英文标注组成，并采用MIT许可。数据集的标注信息包括关系类型、文本token、头实体和尾实体及其类型与索引等。目前，FewRel 支持关系抽取等任务。

FewRel aims to provide large-scale few-shot relation extraction capabilities, containing over one hundred relation types and tens of thousands of annotated instances across different domains. The dataset size of this repository ranges from 10,000 to 100,000 samples, mainly consisting of English text and crowdsourced English annotations, and it is licensed under the MIT License. The annotation information of the dataset includes relation types, text tokens, head entities and tail entities along with their types and indices, etc. Currently, FewRel supports tasks such as relation extraction.