| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Visual Reasoning | Geometric Shapes | Accuracy95.2 | 28 | |
| Poisson Equation Solving | Geometric Shapes Jared, Octopus, Apple, Sofa, Holey Human, Fastener (test) | NRMSE0.0032 | 6 | |
| Geometric Reasoning | Geometric Shapes | Final Accuracy84.5 | 4 | |
| Prompt optimization | Geometric Shapes | Accuracy63.3 | 4 |