Semantic Hijacking on Semantic Hijacking Mutation Pipeline Config B
0FR RateMistral-Nemo
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Mistral-NemoEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 0 | 467 | 467 | 0 | 0 | 33 | |
| Llama-3.1-8BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 2.8 | 498 | 484 | 0 | 14 | 2 | |
| Gemma-3-12BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 7.4 | 500 | 463 | 0 | 37 | 0 | |
| Llama-3.3-70BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 10.4 | 499 | 447 | 0 | 52 | 1 | |
| Mixtral-8x22BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 10.8 | 500 | 446 | 0 | 54 | 0 | |
| Gemma-2-27BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 13.3 | 497 | 431 | 0 | 66 | 3 | |
| Qwen-2.5-72BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 13.4 | 499 | 432 | 0 | 67 | 1 | |
| Mistral-Small-3.2-24BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 24.2 | 499 | 376 | 2 | 121 | 1 | |
| Qwen-3.5-9BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 27.6 | 468 | 338 | 1 | 129 | 32 | |
| GPT-4o-MiniEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 29.2 | 500 | 354 | 0 | 146 | 0 | |
| Llama-3.1-70BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 51.3 | 499 | 243 | 0 | 256 | 1 | |
| Gemma-4-31BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 61.1 | 496 | 193 | 0 | 303 | 4 | |
| Qwen-3.5-27BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 64.2 | 495 | 177 | 0 | 318 | 5 | |
| Qwen-2.5-7BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 70.8 | 497 | 83 | 62 | 352 | 3 | |
| Qwen-3-32BEvaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 78.2 | 490 | 107 | 0 | 383 | 10 | |
| DeepSeek-V3Evaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 89.4 | 500 | 53 | 0 | 447 | 0 | |
| DeepSeek-R1Evaluation Mode=Worker-Only, Configuration=Config B, Sample Size=n=5002026.05 | 90.3 | 496 | 48 | 0 | 448 | 4 |