ResearchBenchmarksCommonsense Question Answering on OpenBookQA 1.0 (test)Commonsense Question Answering on OpenBookQA 1.0 (test)Follow59.4AccuracyGPT-350.66452.93255.257.468Sep 2, 2022Evaluation resultsMethodMethodLinksAccuracyGPT-3Answer predictor=BERTAnswer predictor=BERT2022.0959.4ELABORAnswer predictor=BERT,...Answer predictor=BERT, Elaboration model=GPT2-large2022.0956.4pipelineAnswer predictor=BERT,...Answer predictor=BERT, Elaboration model=GPT2-large2022.0955selftalkAnswer predictor=BERTAnswer predictor=BERT2022.0954.4vanillaAnswer predictor=BERTAnswer predictor=BERT2022.0951