Prompt Classification on SEA-SafeguardBench English
98.9AUPRCSEA-Guard-12B
Evaluation results
| Method | Links | |
|---|---|---|
| SEA-Guard-12Bevaluation_mode=zero-shot2026.02 | 98.9 | |
| ShieldGemma 9Bevaluation_mode=zero-shot2026.02 | 98.6 | |
| Qwen3Guard-Gen 8Bevaluation_mode=zero-shot2026.02 | 98.5 | |
| SEA-Guard-8Bevaluation_mode=zero-shot2026.02 | 98.5 | |
| SEA-Guard-4Bevaluation_mode=zero-shot2026.02 | 98.4 | |
| ShieldGemma 27Bevaluation_mode=zero-shot2026.02 | 97.9 | |
| Qwen3Guard-Gen 4Bevaluation_mode=zero-shot2026.02 | 97.9 | |
| PolyGuard-Ministral 8Bevaluation_mode=zero-shot2026.02 | 97.3 | |
| SEA-Guard-12Bevaluation_mode=zero-shot2026.02 | 95.9 | |
| SEA-Guard-8Bevaluation_mode=zero-shot2026.02 | 95.7 | |
| SEA-Guard-4Bevaluation_mode=zero-shot2026.02 | 95.6 | |
| Qwen3Guard-Gen 8Bevaluation_mode=zero-shot2026.02 | 94.8 | |
| Qwen3Guard-Gen 4Bevaluation_mode=zero-shot2026.02 | 94.1 | |
| LlamaGuard-3 8Bevaluation_mode=zero-shot2026.02 | 93.9 | |
| PolyGuard-Ministral 8Bevaluation_mode=zero-shot2026.02 | 93.8 | |
| LlamaGuard-3 8Bevaluation_mode=zero-shot2026.02 | 90.8 | |
| ShieldGemma 27Bevaluation_mode=zero-shot2026.02 | 86 | |
| ShieldGemma 9Bevaluation_mode=zero-shot2026.02 | 85 |