Adversarial Intent Detection on trajectory window 10-round
82Accuracyo3
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| o3Model=o3, Reasoning=with reasoning2026.03 | 82 | 0.001 | |
| gpt-4.1Model=gpt-4.1, Reasoning=with reasoning2026.03 | 81.2 | 0.001 | |
| gpt-4.1-miniModel=gpt-4.1-mini, Reasoning=with reasoning2026.03 | 69.4 | 0.001 | |
| gpt-4oModel=gpt-4o, Reasoning=with reasoning2026.03 | 51.2 | 0.311 |