Reasoning & Coding on WaterBench (test)
59.82GMAAR
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AARModel=OPT-1.3B2025.12 | 59.82 | 0.78 | 0.695 | |
| OriginalModel=OPT-1.3B2025.12 | 51.19 | — | — | |
| SynthIDModel=OPT-1.3B2025.12 | 50.85 | 0.985 | 0.545 | |
| SWEETModel=OPT-1.3B2025.12 | 50.04 | 0.825 | 0.88 | |
| DualGuardModel=OPT-1.3B2025.12 | 49.88 | 0.94 | 0.655 | |
| UnbiasedModel=OPT-1.3B2025.12 | 49.4 | 0.81 | 0.64 | |
| DIPmarkModel=OPT-1.3B2025.12 | 49.27 | 0.935 | 0.5 | |
| EWDModel=OPT-1.3B2025.12 | 49.24 | 0.835 | 0.935 | |
| KGWModel=OPT-1.3B2025.12 | 48.84 | 0.885 | 0.8 | |
| XSIRModel=OPT-1.3B2025.12 | 48.7 | 0.81 | 0.61 | |
| SIRModel=OPT-1.3B2025.12 | 47.1 | 0.97 | 0.735 |