OOD Safety Category Inference (Stage 2) on Aegis 2.0
23.08Reward MeanProGuard-3B
Evaluation Results
| Method | Links | |
|---|---|---|
| ProGuard-3B2025.12 | 23.08 | |
| ProGuard-7B2025.12 | 20.28 | |
| Gemini2.5-Flash2025.12 | 15.84 | |
| GPT4o-mini2025.12 | 0.61 |
Loading the SOTA2 catalog…
| Method | Links | |
|---|---|---|
| ProGuard-3B2025.12 | 23.08 | |
| ProGuard-7B2025.12 | 20.28 | |
| Gemini2.5-Flash2025.12 | 15.84 | |
| GPT4o-mini2025.12 | 0.61 |