ReCoRD is the abbreviation for Reading Comprehension with Commonsense Reasoning Dataset. It is used to evaluate whether language models can infer masked entities from passages using contextual and commonsense reasoning.
Abbreviation for Reading Comprehension with Commonsense Reasoning Dataset.