Sorrybench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
SorryBench
8.9ASR (SorryBench)
62
SorryBench
54.5Reasoning Success Rate (FFR)
32
SorryBench
4.2Harmful Response Rate (%)
18
SorryBench
56.8FFR (Reasoning)
17
SorryBench
89.43F1 Score
14
SorryBench
43.6LG4 ASR
8
Sorrybench
12.95Harmful Score
8
SorryBench
8.22ASR
6
SorryBench wrapped with HarmBench templates (held-out)
86.9Detection Rate
3
SorryBench clean condition (held-out)
95Detection Rate
3