Geometric Reasoning on Geometric Shapes
84.5Final AccuracyTRAS**
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TRAS**Configuration=MCSA + TR2025.07 | 84.5 | 0.008 | 2.175 | 54.8 | 58.2 | |
| Baseline + TR*Regularization=Textual Regularization from iter. 02025.07 | 81.7 | 0.021 | 1.641 | 54.8 | 58.2 | |
| Baseline + MCSA*Aggregation samples (K)=22025.07 | 79.4 | 0.016 | 1.782 | 54.8 | 58.2 | |
| PromptAgentstatus=Baseline2025.07 | 75.1 | — | — | 54.8 | 58.2 |