Liars' Bench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Liars' Bench Harm-Pressure Knowledge (test)
0.91AUROC
3
Liars' Bench Insider Trading (test)
0.953AUROC
3
Liars' Bench Convincing Game (test)
1AUROC
3
Liars' Bench Harm-Pressure Choice (test)
0.949AUROC
3
Liars' Bench Instructed Deception (test)
0.939AUROC
3