Financial Compliance Classification on FinGuard-Bench query-level
90.23F1 ScoreFinGuard
Evaluation Results
| Method | Links | |
|---|---|---|
| FinGuardself-play=true2026.05 | 90.23 | |
| FinGuardself-play=false2026.05 | 84.13 | |
| Qwen3.5-397B-A17B2026.05 | 78.68 | |
| GPT-5.12026.05 | 74.26 | |
| Gemini-3.1-Flash2026.05 | 61.28 | |
| PolyGuard-Qwen-7B2026.05 | 52.46 | |
| Qwen3Guard-Gen-8Bmode=strict2026.05 | 35.5 | |
| Qwen3.5-9B2026.05 | 33.63 | |
| LlamaGuard3-8B2026.05 | 23.94 | |
| NemoGuard-8B2026.05 | 16.97 | |
| Qwen3-8B2026.05 | 16.69 | |
| Qwen3Guard-Gen-8Bmode=loose2026.05 | 11.16 |