ResearchBenchmarksAdversarial Attack on AdvBench unsafe inputsFollow94ASRQwen 2.519.1238.565877.44Jan 27, 2025Evaluation ResultsMethodMethodLinksASRQwen 2.5Attack Type=White-box,...Attack Type=White-box, Input Setting=Unsafe inputs2025.0194GraniteAttack Type=White-box,...Attack Type=White-box, Input Setting=Unsafe inputs2025.0191Gemma 1Attack Type=White-box,...Attack Type=White-box, Input Setting=Unsafe inputs2025.0162Llama 2Attack Type=White-box,...Attack Type=White-box, Input Setting=Unsafe inputs2025.0122