Malicious Action Detection on tau-Bench retail (test)
7.8TPRGPT-OSS-Safeguard 20B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-OSS-Safeguard 20B2025.10 | 7.8 | 51.8 | |
| AprielGuard2025.10 | 4.3 | 6.3 | |
| Granite-Guardian 3.3-8B2025.10 | 0 | 0.9 | |
| Qwen-3-Guard2025.10 | 0 | 0.2 |
| Method | Links | ||
|---|---|---|---|
| GPT-OSS-Safeguard 20B2025.10 | 7.8 | 51.8 | |
| AprielGuard2025.10 | 4.3 | 6.3 | |
| Granite-Guardian 3.3-8B2025.10 | 0 | 0.9 | |
| Qwen-3-Guard2025.10 | 0 | 0.2 |