TriviaQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
TriviaQA
76.4Token-Level F1
20
TriviaQA
81.6ECR
20
TriviaQA
74.4EM
20
TriviaQA
71.58Exact Match
20
TriviaQA-generated Fact Dataset topic-wise leave-one-out strategy 1.0
63.9Accuracy (billturnbull)
20
TriviaQA
81.4Accuracy
20
TriviaQA (out-of-domain)
68Accuracy
19
TriviaQA 10K context
80.8Accuracy
19
TriviaQA DPR
72.6EM
19
TriviaQA adversarial Contriever retrieved top 10
57.72EM
19
TriviaQA
49Macro EM
19
TriviaQA Wikipedia (dev-full)
61.3F1
19
TriviaQA 128K
70Accuracy
18
TQA (TriviaQA) (test)
68.4F1 Score
18
TriviaQA
22.4PRR
18
TriviaQA
18.5ΔLL (%)
18
TriviaQA
82.8Pass@1
18
TriviaQA (in-domain)
3.4Expected Calibration Error (ECE)
18
TriviaQA
88.07F1 Score
18
TriviaQA
90.6Response Accuracy
18
TriviaQA
80.2Accuracy
18
TriviaQA
37.74EM
18
TriviaQA Generalization
80F1
18
TriviaQA
64.4EM
18
TriviaQA 100K context
70.2Accuracy
18