QA Tasks
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
5 QA tasks
54.02Accuracy
78
7 QA tasks
69.44Accuracy
42
QA Tasks (unseen)
42.12AN' Score
14
QA Tasks seen (test)
59.23AN
14
QA Tasks (OpenBookQA, PIQA, ARC-E, ARC-C, SciQ, WebQs) (test)
0.394OpenBookQA
12
Seen QA Tasks
79.99SQuAD 2
8