Inference Attack on SVAMP
97.61AUCCIA
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CIAMAS Framework=ARG-Designer2026.04 | 97.61 | 91.06 | 86.32 | |
| CIAMAS Framework=AGP2026.04 | 89.79 | 83.8 | 79.23 | |
| CIAMAS Framework=G-Designer2026.04 | 85.61 | 80.83 | 76.82 | |
| Gemini-2.5-ProMAS Framework=G-Designer2026.04 | 63 | 76 | 70.59 | |
| GPT-5MAS Framework=G-Designer2026.04 | 62.72 | 78.76 | 68.15 | |
| GPT-5MAS Framework=ARG-Designer2026.04 | 62.4 | 71.63 | 40.34 | |
| GPT-5MAS Framework=AGP2026.04 | 61.99 | 79.4 | 63.3 | |
| Gemini-2.5-ProMAS Framework=ARG-Designer2026.04 | 61.97 | 73.13 | 40.97 | |
| Gemini-2.5-ProMAS Framework=AGP2026.04 | 60.87 | 77.33 | 62.41 | |
| Llama-3.1-8B-InstructMAS Framework=ARG-Designer2026.04 | 53.49 | 62.77 | 43.52 | |
| Mistral-7B-Instruct-v0.2MAS Framework=G-Designer2026.04 | 49.77 | 67.04 | 43.51 | |
| Mistral-7B-Instruct-v0.2MAS Framework=AGP2026.04 | 49.39 | 65.88 | 44.75 | |
| Llama-3.1-8B-InstructMAS Framework=G-Designer2026.04 | 45.67 | 62.52 | 43.61 | |
| Llama-3.1-8B-InstructMAS Framework=AGP2026.04 | 43.55 | 63.15 | 53.96 | |
| Mistral-7B-Instruct-v0.2MAS Framework=ARG-Designer2026.04 | 31.12 | 53.47 | 12.35 |