HarmBench and AdvBench
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Jailbreak Defense | HarmBench and AdvBench (test) | GCG Score91.2 | 44 | |
| LLM Inference efficiency | HarmBench-Standard and AdvBench | Slowdown2 | 8 | |
| Generative AI Output Safety | HarmBench and AdvBench (test) | Safe Rate82.88 | 8 |
Showing 3 of 3 rows