TriviaQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
TriviaQA (TQA) standard (test)
60.3Exact Match
10
TriviaQA Open (eval)
76.94F1 Score
9
TriviaQA (official)
64.7EM
9
TriviaQA In-Distribution
96.09Exact Match (EM)
9
TriviaQA (1,000-example evaluation slice)
25.6Alias-Substring Accuracy
9
TriviaQA
83.7EM
9
TriviaQA (out-of-distribution)
80.44AUC
9
TriviaQA
62.2Exact Match (EM)
9
TriviaQA
87AUROC
9
TriviaQA
1.408PPL
9
TriviaQA
72.6Recall@5
9
TriviaQA
79.4Recall@20
9
TriviaQA Open-domain filtered (test)
79.8Exact Match (EM)
9
TriviaQA
84.1Recall@20
9
TriviaQA hidden (test)
80.1EM
9
TriviaQA
59.65EM
8
TriviaQA
71.7Pass@1 Accuracy
8
TriviaQA
93.18CF
8
TriviaQA
75.3F1 Score
8
TriviaQA (TQA) 2048 tokens
68.2Accuracy (TQA)
8
TriviaQA (Held-out)
82.2Accuracy
8
TriviaQA Commonsense
69.6Accuracy
8
TriviaQA
59.36Exact Match (EM)
8
TriviaQA
2.718ECE' Gain
8
TriviaQA
63.21F1 Score
8