SQuAD

Short for the Stanford Question Answering Dataset, a benchmark dataset used to evaluate machine reading comprehension and question answering systems. SQuAD became historically important because it provided a widely adopted way to compare progress in extractive question answering, especially in the pre-LLM era of NLP research.