Two is Better than Many? Binary Classification as an Effective Approach to Multi-Choice Question Answering

Benchmark Model Rank Results
question-answering-on-piqaDeBERTa-Large 304M#5Accuracy: 87.4
question-answering-on-piqaDeBERTa-Large 304M (classification-based)#8Accuracy: 85.9
question-answering-on-social-iqaDeBERTa-Large 304M#6Accuracy: 80.2
question-answering-on-social-iqaDeBERTa-Large 304M (classification-based)#7Accuracy: 79.9
sentence-completion-on-hellaswagDeBERTa-Large 304M (classification-based)#3Accuracy: 95.6
sentence-completion-on-hellaswagDeBERTa-Large 304M#6Accuracy: 94.7