ResearchBenchmarksJailbreak Defense on StrongREJECT AutoDAN-Turbo attack (test)Follow87LlamaGuard ScoreReSA-SFT52.6861.5970.579.41Sep 15, 2025Evaluation ResultsMethodMethodLinksLlamaGuard ScoreStrongREJECT Evaluator ScoreHarmBench Classifier ScoreReSA-SFTBase Model=Qwen2.5-7B-...Base Model=Qwen2.5-7B-Instruct, Jailbreak=AutoDAN-Turbo2025.098787.3396WJ-SFTBase Model=Qwen2.5-7B-...Base Model=Qwen2.5-7B-Instruct, Jailbreak=AutoDAN-Turbo2025.096678.9293BaseBase Model=Qwen2.5-7B-...Base Model=Qwen2.5-7B-Instruct, Jailbreak=AutoDAN-Turbo2025.095458.5679