Share your thoughts, 1 month free Claude Pro on usSee more

Jailbreak Defense on WordGame

14.65ASR

GuardReasoner-1B

Updated 1mo ago

Evaluation Results

Method	Links
GuardReasoner-1B 2026.06		14.65
YuFeng-XGuard-0.6B Prompt+Response 2026.06		22.29
w/o Future-Aware Reasoning 2026.06		25.48
FreoStream 2026.06		27.39
YuFeng-XGuard-0.6B Prompt Only 2026.06		35.03
Qwen-Stream-8B 2026.06		35.03
w/o Safety-Aligned Optimization 2026.06		35.03
Qwen-Stream-0.6B 2026.06		36.31
Nemotron-4B 2026.06		50.96
Qwen-Gen-0.6B Prompt Only 2026.06		68.79
Qwen-Gen-0.6B Prompt+Response 2026.06		73.25