List your company

Jailbreaking on AdvBench (test)

99.04Average ASRCHaRS-PCT
23.785643.322862.8682.3972May 22, 2025Jul 5, 2025Aug 19, 2025Oct 2, 2025Nov 16, 2025Dec 30, 2025Feb 13, 2026

Evaluation results

MethodLinks
CHaRS-PCTBackbone=Llama3.1-8B-Instruct, Intervention protocol=ActAdd2026.0299.04——————
CHaRSBackbone=Gemma2-9B-Instruct, Intervention protocol=ActAdd2026.0298.08——————
CHaRS-PCTBackbone=Gemma2-9B-Instruct, Intervention protocol=ActAdd2026.0298.08——————
CHaRSBackbone=Llama3.1-8B-Instruct, Intervention protocol=ActAdd2026.0298.08——————
ActAddBackbone=Llama3.1-8B-Instruct, Intervention protocol=ActAdd2026.0295.19——————
CHaRSBackbone=Qwen2.5-3B-Instruct, Intervention protocol=ActAdd2026.0295.19——————
CHaRS-PCTBackbone=Qwen2.5-3B-Instruct, Intervention protocol=ActAdd2026.0295.19——————
CHaRSBackbone=Qwen2.5-7B-Instruct, Intervention protocol=ActAdd2026.0295.19——————
CHaRSBackbone=Qwen2.5-14B-Instruct, Intervention protocol=ActAdd2026.0295.19——————
ActAddBackbone=Qwen2.5-14B-Instruct, Intervention protocol=ActAdd2026.0294.23——————
CHaRS-PCTBackbone=Qwen2.5-7B-Instruct, Intervention protocol=ActAdd2026.0293.27——————
CHaRS-PCTBackbone=Qwen2.5-14B-Instruct, Intervention protocol=ActAdd2026.0293.27——————
ActAddBackbone=Gemma2-9B-Instruct, Intervention protocol=ActAdd2026.0291.35——————
ActAddBackbone=Qwen2.5-7B-Instruct, Intervention protocol=ActAdd2026.0291.35——————
ActAddBackbone=Qwen2.5-3B-Instruct, Intervention protocol=ActAdd2026.0289.42——————
CHaRSBackbone=Qwen2.5-32B-Instruct, Intervention protocol=ActAdd2026.0289.42——————
CHaRS-PCTBackbone=Qwen2.5-32B-Instruct, Intervention protocol=ActAdd2026.0288.46——————
ActAddBackbone=Qwen2.5-32B-Instruct, Intervention protocol=ActAdd2026.0282.69——————
CHaRSBackbone=Llama3.2-3B-Instruct, Intervention protocol=ActAdd2026.0279.81——————
CHaRS-PCTBackbone=Llama3.2-3B-Instruct, Intervention protocol=ActAdd2026.0279.81——————
ActAddBackbone=Llama3.2-3B-Instruct, Intervention protocol=ActAdd2026.0274.04——————
MTSA-R3iteration=32025.0563.927266.55650.56474.5
MTSA-R2iteration=22025.0561.168.564.556.547.558.571
MTSA-R1iteration=12025.0558.0864.56554.546.55167
RedQueen2025.0553.426358.55343.54557.5
COA2025.05515263.55524.54863
PAIR2025.0548.557.56151.520.539.561
ActorAttack2025.0543.9247.551.544.53541.543.5
CodeAttack2025.0543.7544.55638.53743.543
CipherChat2025.0532.3341.53228.5223832
GCG2025.0527.1733.512.52234.524.536
PAP2025.0526.913624.514.52632.528
HarmfulQA2025.0526.6834.5393316.52116