Adversarial Code Compliance on Overall Mean
97.1Decoupling ProbabilityLlama-3.1-8B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Llama-3.1-8BModel Category=Open Source (Small), Strategy=RPA^C2026.01 | 97.1 | 48.4 | |
| DeepSeek-v3.2Model Category=Open Source (Large), Strategy=RPA^C2026.01 | 95.8 | 51 | |
| Gemma-3-27bModel Category=Open Source (Small), Strategy=RPA^C2026.01 | 92.2 | 38 | |
| Qwen3-235BModel Category=Open Source (Large), Strategy=RPA^C2026.01 | 89 | 30.4 | |
| Gemini-2.5-FlashModel Category=Proprietary, Strategy=RPA^C2026.01 | 83.6 | 42.9 | |
| Llama-3.2-3BModel Category=Open Source (Small), Strategy=RPA^C2026.01 | 77.7 | 31.7 | |
| GPT-5Model Category=Proprietary, Strategy=RPA^C2026.01 | 71.6 | 28.3 | |
| GPT-OSS-120BModel Category=Open Source (Large), Strategy=RPA^C2026.01 | 32.4 | 1.6 | |
| GPT-5-MiniModel Category=Proprietary, Strategy=RPA^C2026.01 | 29.5 | 0.6 |