Also known as: Reading Comprehension With Commonsense Reasoning Dataset (ReCoRD)
Reading Comprehension with Commonsense Reasoning Dataset, or ReCoRD, is a benchmark for evaluating whether a language model can use commonsense reasoning to answer masked-entity questions from passages. It is part of the SuperGLUE benchmark suite and tests whether a model can infer the correct entity from contextual evidence.
A dataset to evaluate an LLM's ability to perform commonsense reasoning. Each example in the dataset contains three components: - A paragraph or two from a news article - A query in which one of the entities explicitly or implicitly identified in the passage is masked. - The answer (the name of the entity that belongs in the mask) See ReCoRD for an extensive list of examples. ReCoRD is a component of the SuperGLUE ensemble.