Jailbreak Defense on ReNeLLM
0ASRIA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| IAModel=GPT-4o-mini2025.05 | 0 | 1 | |
| SAGEModel=GPT-4o-mini2025.05 | 0 | 1 | |
| Goal PrioritizationModel=GPT-4o2025.05 | 0 | 1 | |
| SAGEModel=GPT-4o2025.05 | 0 | 1 | |
| ICDModel=Claude-3.5-Sonnet2025.05 | 0 | 1.02 | |
| Goal PrioritizationModel=Claude-3.5-Sonnet2025.05 | 0 | 1 | |
| SAGEModel=Claude-3.5-Sonnet2025.05 | 0 | 1 | |
| SAGEModel=Gemma22025.05 | 0 | 1 | |
| IAModel=Llama3.12025.05 | 0 | 1 | |
| IAModel=GPT-4o2025.05 | 0.04 | 1.08 | |
| Self-ReminderModel=Claude-3.5-Sonnet2025.05 | 0.06 | 1.08 | |
| Self-ExaminationModel=GPT-4o-mini2025.05 | 0.12 | 1.34 | |
| IAModel=Claude-3.5-Sonnet2025.05 | 0.14 | 1.06 | |
| Self-ExaminationModel=Claude-3.5-Sonnet2025.05 | 0.16 | 1.62 | |
| No DefenseModel=Claude-3.5-Sonnet2025.05 | 0.2 | 1.72 | |
| Self-ExaminationModel=GPT-4o2025.05 | 0.22 | 1.76 | |
| Goal PrioritizationModel=GPT-4o-mini2025.05 | 0.24 | 1.64 | |
| ICDModel=GPT-4o2025.05 | 0.24 | 1 | |
| ICDModel=GPT-4o-mini2025.05 | 0.36 | 2.32 | |
| Self-ReminderModel=GPT-4o-mini2025.05 | 0.52 | 2.02 | |
| Self-ReminderModel=GPT-4o2025.05 | 0.6 | 1.7 | |
| No DefenseModel=GPT-4o2025.05 | 0.98 | 4.74 | |
| No DefenseModel=GPT-4o-mini2025.05 | 1 | 4.74 | |
| Self-ExaminationModel=Llama3.12025.05 | 2 | 1.08 | |
| SAGEModel=Llama3.12025.05 | 2 | 1 | |
| SAGEModel=Qwen2.52025.05 | 6 | 1.16 | |
| Goal PrioritizationModel=Qwen2.52025.05 | 8 | 1.24 | |
| ICDModel=Llama3.12025.05 | 10 | 1.4 | |
| FreoStreamGuardrail Type=Stream, Protected LLM=GPT-4o, Future-Aware Reasoning=true, Safety-Aligned Optimization=true2026.06 | 10.1 | — | |
| w/o Future-Aware ReasoningGuardrail Type=Stream, Protected LLM=GPT-4o, Future-Aware Reasoning=false, Safety-Aligned Optimization=true2026.06 | 10.1 | — | |
| Qwen-Stream-0.6BGuardrail Type=Stream, Protected LLM=GPT-4o2026.06 | 11.49 | — | |
| w/o Safety-Aligned OptimizationGuardrail Type=Stream, Protected LLM=GPT-4o, Future-Aware Reasoning=true, Safety-Aligned Optimization=false2026.06 | 11.86 | — | |
| Nemotron-4BGuardrail Type=Generative, Protected LLM=GPT-4o2026.06 | 12.28 | — | |
| Qwen-Gen-0.6B Prompt+ResponseGuardrail Type=Generative, Input Context=Prompt+Response, Protected LLM=GPT-4o2026.06 | 16.83 | — | |
| Goal PrioritizationModel=Gemma22025.05 | 18 | 1.2 | |
| Qwen-Stream-8BGuardrail Type=Stream, Protected LLM=GPT-4o2026.06 | 22.01 | — | |
| YuFeng-XGuard-0.6B Prompt OnlyGuardrail Type=Generative, Input Context=Prompt Only, Protected LLM=GPT-4o2026.06 | 23.56 | — | |
| YuFeng-XGuard-0.6B Prompt+ResponseGuardrail Type=Generative, Input Context=Prompt+Response, Protected LLM=GPT-4o2026.06 | 25.76 | — | |
| Self-ExaminationModel=Qwen2.52025.05 | 28 | 2.02 | |
| Self-ReminderModel=Llama3.12025.05 | 28 | 1.64 | |
| GuardReasoner-1BGuardrail Type=Generative, Protected LLM=GPT-4o2026.06 | 28.71 | — | |
| Self-ExaminationModel=Gemma22025.05 | 38 | 1.58 | |
| Goal PrioritizationModel=Llama3.12025.05 | 50 | 1.54 | |
| No DefenseModel=Llama3.12025.05 | 70 | 3.68 | |
| Qwen-Gen-0.6B Prompt OnlyGuardrail Type=Generative, Input Context=Prompt Only, Protected LLM=GPT-4o2026.06 | 81.19 | — | |
| ICDModel=Qwen2.52025.05 | 88 | 4.2 | |
| No DefenseModel=Gemma22025.05 | 96 | 4.76 | |
| ICDModel=Gemma22025.05 | 96 | 4.56 | |
| Self-ReminderModel=Gemma22025.05 | 98 | 3.72 | |
| IAModel=Gemma22025.05 | 98 | 3.14 | |
| Self-ReminderModel=Qwen2.52025.05 | 98 | 4.38 | |
| No DefenseModel=Qwen2.52025.05 | 100 | 4.86 | |
| IAModel=Qwen2.52025.05 | 100 | 4.3 |