SIR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
SIR
0.066Area of HPD region
10
SIR(2D) ID
0Residual NMSE
10
SIR 2D ID-Ext
0Residual NMSE
8
SIR delay structural-permutation (test)
100Gap Closed toward Oracle
7
SIR
99.6Attack Success Rate
6
SIR
-503.37Avg NLL (Data | Model)
6
SIR
0.058RMSE
5
SIR high-dimensional (test)
0.76C2ST Accuracy
5
SIR Scaled domain (test)
0.18NLL
4
SIR Shifted domain (test)
0.19NLL
4
SIR Original domain (test)
0.19NLL
4
SIR tweaked
20F1 Score
3
SIR OOD-T3 tweaked (test)
0.001MSE
3
SIR OOD-T2 tweaked (test)
2.4MSE (1e-4)
3
sir distractors
0.645Joint C2ST
3
sir
0.647Joint C2ST
3
sir distractors
0.599Joint C2ST
3
SIR
100Success Rate
3
SIR epsilon=0.05
100Success Fraction
3
SIR undercounting contamination model
—Primary metric
0