Inference Attack on GSM8K
0.9873AUCCIA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CIAMAS Framework=ARG-Designer2026.04 | 0.9873 | 90.35 | 83.3 | |
| CIAMAS Framework=AGP2026.04 | 0.8804 | 82.04 | 78.66 | |
| CIAMAS Framework=G-Designer2026.04 | 0.8585 | 83.82 | 77.44 | |
| Gemini-2.5-ProMAS Framework=ARG-Designer2026.04 | 0.7475 | 79.77 | 57.43 | |
| GPT-5MAS Framework=ARG-Designer2026.04 | 0.6984 | 76.72 | 51.13 | |
| GPT-5MAS Framework=G-Designer2026.04 | 0.6274 | 58.87 | 45.55 | |
| GPT-5MAS Framework=AGP2026.04 | 0.6233 | 59.72 | 53.57 | |
| Gemini-2.5-ProMAS Framework=AGP2026.04 | 0.6177 | 57.33 | 51.74 | |
| Gemini-2.5-ProMAS Framework=G-Designer2026.04 | 0.6152 | 57.28 | 45.45 | |
| Mistral-7B-Instruct-v0.2MAS Framework=ARG-Designer2026.04 | 0.5713 | 66.97 | 49.24 | |
| Llama-3.1-8B-InstructMAS Framework=G-Designer2026.04 | 0.5284 | 53.48 | 46.89 | |
| Llama-3.1-8B-InstructMAS Framework=ARG-Designer2026.04 | 0.5123 | 73.45 | 64.35 | |
| Mistral-7B-Instruct-v0.2MAS Framework=AGP2026.04 | 0.4867 | 65 | 28.66 | |
| Llama-3.1-8B-InstructMAS Framework=AGP2026.04 | 0.4265 | 51.9 | 51 | |
| Mistral-7B-Instruct-v0.2MAS Framework=G-Designer2026.04 | 0.4105 | 42.6 | 28.66 |