Adversarial Attack Misdirection on MTJ-Pro (test)
60.9Purpose Reverse (MSR)HoneyTrap
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| HoneyTrapLLM=LLaMa2026.01 | 60.9 | 75 | 28.6 | 57.9 | 55.6 | 70 | 36.8 | 54 | |
| HoneyTrapLLM=GPT-42026.01 | 52.2 | 50 | 71.4 | 73.7 | 33.3 | 60 | 73.7 | 59 | |
| HoneyTrapLLM=Gemini2026.01 | 52.2 | 75 | 57.1 | 94.7 | 61.1 | 70 | 63.2 | 67 | |
| Self-ReminderLLM=LLaMa2026.01 | 43.5 | 0 | 28.6 | 0 | 11.1 | 50 | 15.8 | 28.6 | |
| Self-ReminderLLM=Gemini2026.01 | 39.1 | 25 | 28.6 | 26.3 | 22.2 | 50 | 21.1 | 32 | |
| PATLLM=GPT-42026.01 | 34.8 | 25 | 28.6 | 26.3 | 16.7 | 20 | 15.8 | 27.6 | |
| PATLLM=LLaMa2026.01 | 34.8 | 0 | 42.9 | 26.3 | 33.3 | 30 | 15.8 | 28.6 | |
| RPOLLM=GPT-42026.01 | 34.8 | 25 | 14.3 | 31.6 | 5.6 | 10 | 10.5 | 20.7 | |
| RPOLLM=Gemini2026.01 | 34.8 | 0 | 28.6 | 15.8 | 16.7 | 20 | 0 | 20.3 | |
| HoneyTrapLLM=GPT-3.52026.01 | 34.8 | 50 | 42.9 | 73.7 | 55.6 | 70 | 47.4 | 53 | |
| PATLLM=GPT-3.52026.01 | 17.4 | 0 | 0 | 5.3 | 5.6 | 0 | 0 | 9.9 | |
| PATLLM=Gemini2026.01 | 17.4 | 25 | 0 | 15.8 | 27.8 | 20 | 21.1 | 18 | |
| RPOLLM=GPT-3.52026.01 | 17.4 | 0 | 0 | 0 | 0 | 0 | 0 | 9.9 | |
| RPOLLM=LLaMa2026.01 | 17.4 | 0 | 28.6 | 26.3 | 27.8 | 50 | 5.3 | 21 | |
| Self-ReminderLLM=GPT-3.52026.01 | 8.7 | 75 | 0 | 10.5 | 16.7 | 20 | 21.1 | 24.3 | |
| Self-ReminderLLM=GPT-42026.01 | 8.7 | 50 | 0 | 10.5 | 22.2 | 10 | 31.6 | 20.7 | |
| GoalPriorityLLM=GPT-3.52026.01 | 8.7 | 0 | 0 | 15.8 | 5.6 | 0 | 10.5 | 5.7 | |
| GoalPriorityLLM=LLaMa2026.01 | 8.7 | 0 | 0 | 10.5 | 5.6 | 10 | 15.8 | 6.4 | |
| GoalPriorityLLM=GPT-42026.01 | 4.3 | 0 | 0 | 5.3 | 5.6 | 0 | 15.8 | 5.9 | |
| GoalPriorityLLM=Gemini2026.01 | 0 | 0 | 0 | 0 | 0 | 0 | 5.3 | 1.5 |