Input Moderation on WildGuard (test)
89.5F1 ScoreLlama3-StreamGuard-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| Llama3-StreamGuard-8BStreaming=yes2026.04 | 89.5 | |
| Llama3-StreamGuard-3BStreaming=yes2026.04 | 89 | |
| WildGuard-7BStreaming=no2026.04 | 88.9 | |
| Qwen3Guard-8B-Gen-strictStreaming=no2026.04 | 88.9 | |
| Qwen3Guard-Stream-4B-strictStreaming=yes2026.04 | 88.6 | |
| Llama3-StreamGuard-1BStreaming=yes2026.04 | 88.6 | |
| PolyGuard-Qwen-7BStreaming=no2026.04 | 88.1 | |
| Qwen3Guard-0.6B-Gen-strictStreaming=no2026.04 | 87.7 | |
| Qwen3Guard-Stream-8B-strictStreaming=yes2026.04 | 87.5 | |
| Qwen3Guard-Stream-0.6B-strictStreaming=yes2026.04 | 87.1 | |
| Qwen3Guard-Stream-0.6B-looseStreaming=yes2026.04 | 86 | |
| Qwen3Guard-4B-Gen-strictStreaming=no2026.04 | 85.6 | |
| Qwen3Guard-8B-Gen-looseStreaming=no2026.04 | 85.6 | |
| Qwen3Guard-Stream-4B-looseStreaming=yes2026.04 | 85.3 | |
| Qwen3Guard-0.6B-Gen-looseStreaming=no2026.04 | 85.1 | |
| Qwen3Guard-4B-Gen-looseStreaming=no2026.04 | 85.1 | |
| Qwen3Guard-Stream-8B-looseStreaming=yes2026.04 | 84.4 | |
| NemoGuard-8BStreaming=no2026.04 | 81.6 | |
| LlamaGuard3-8BStreaming=no2026.04 | 76.4 | |
| LlamaGuard4-12BStreaming=no2026.04 | 73 | |
| ShieldGemma-27BStreaming=no2026.04 | 54.3 | |
| ShieldGemma-9BStreaming=no2026.04 | 54.2 |