List your company

Jailbreak Attack on HarmBench standard behaviors

61ASRAdvGRPO
59.967.32574.7582.175Jun 8, 2026

Evaluation results

MethodLinks
AdvGRPOTransfer Defender=Llama-3.1-8B, Attacker=Qwen2.5-14B, Turn Type=Single-turn (ST)2026.0661
AdvGRPOTransfer Defender=Phi-4-mini, Attacker=Qwen2.5-14B, Turn Type=Single-turn (ST)2026.0667.5
AdvGRPOTransfer Defender=Gemma-2-9B, Attacker=Qwen2.5-14B, Turn Type=Single-turn (ST)2026.0668.5
SEMATransfer Defender=Phi-4-mini, Attacker=Qwen2.5-14B2026.0671.5
SEMATransfer Defender=Gemma-2-9B, Attacker=Qwen2.5-14B2026.0675.5
SEMATransfer Defender=Llama-3.1-8B, Attacker=Qwen2.5-14B2026.0679.5
AdvGRPOTransfer Defender=Phi-4-mini, Attacker=Qwen2.5-14B, Turn Type=Multi-turn (MT)2026.0682
AdvGRPOTransfer Defender=Gemma-2-9B, Attacker=Qwen2.5-14B, Turn Type=Multi-turn (MT)2026.0686
AdvGRPOTransfer Defender=Llama-3.1-8B, Attacker=Qwen2.5-14B, Turn Type=Multi-turn (MT)2026.0688.5