ResearchBenchmarksQuestion Answering on ELI5 (unseen test)Follow13.97ROUGE-1QA_s12.316412.745713.17513.6043Oct 13, 2022Evaluation ResultsMethodMethodLinksROUGE-1ROUGE-2ROUGE-LROUGE-LsumQA_sDescription=BART-base...Description=BART-base pre-trained on 200K synthetic QA pairs generated by the proposed best QG model, Mode=Unsupervised2022.1013.972.291011.77QA_bDescription=BART-base...Description=BART-base model without any synthetic pre-training, Mode=Unsupervised2022.1012.382.198.9810.63