TriviaQA, HotpotQA, and MedMCQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
TriviaQA, HotpotQA, and MedMCQA In-Distribution Average (val)
0.91Head Entropy
12
TriviaQA, HotpotQA, and MedMCQA OOD Generalization Average (val)
0.68Head Entropy
6