PrefEval
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
PrefEval
77.96Accuracy
27
PrefEval
99.99Accuracy
12
PrefEval 10 injected adversarial turns
7.4Pref Unaware Rate
10
PrefEval Implicit
69.9Accuracy
8
PrefEval Explicit
81.3Accuracy
8
PrefEval
68.8AccPF
7
PrefEval Implicit Preference (test)
19.1Score
6
PrefEval Implicit Choice (test)
78.6Choice Accuracy
6
PrefEval Explicit Preference (test)
77.7LLM-Evaluated Score
6
PrefEval (test)
38.33Accuracy
4