SPHERE
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Function Optimization | Sphere D=1000 | Final Value3.9387 | 19 | |
| High-dimensional optimization | Sphere D=10000 | Objective Value (Sphere D=10000)6.4338 | 13 | |
| Named Entity Recognition | SPHERE | CS NER Score80.47 | 12 | |
| Black-box Optimization | Sphere d=20 | Objective Value (Median)0.0032 | 9 | |
| Pose Graph Optimization | Sphere | Objective Value1.687 | 9 | |
| Classification | Sphere (test) | Accuracy100 | 9 | |
| Continuous Black-box Optimization | Sphere 30D | Average Objective Value0 | 7 | |
| Heat Equation Solving | Sphere analytic GT | NRMSE0 | 6 | |
| Relation Extraction | SPHERE Material Science | Hierarchical F181.91 | 5 | |
| Relation Extraction | SPHERE Biology | Rel+ F1 (Hierarchical)82.53 | 5 | |
| Relation Extraction | SPHERE Physics | Hierarchical Relation F176.93 | 5 | |
| Relation Extraction | SPHERE Computer Science | Rel+ F1 (Hierarchical)77.4 | 5 | |
| Intrinsic Dimension Estimation | Sphere d = 1 | Estimated Intrinsic Dimension1.94 | 4 | |
| Spatial Reasoning | SPHERE | Single Skill Performance71.2 | 3 | |
| Relation Extraction | SPHERE Mat. Sci. 1.0 (test) | F1 (Hierarchical)69.92 | 3 | |
| Relation Extraction | SPHERE Comp. Sci. 1.0 (test) | Rel+ F1 (Hierarchical)59.36 | 3 | |
| Intrinsic Dimension Estimation | Sphere d = 5 | Estimated Intrinsic Dimension5.13 | 2 | |
| Geodesic Pose Graph Optimization | Sphere | Cost6.9539 | 2 | |
| Sub-pixel Localization | SPHERE SHINE-F150 (test) | RMSE (Position Error)0.11 | 2 | |
| Flux Estimation | SPHERE SHINE-F150 (test) | ARE (Flux Error)51 | 2 | |
| Multi-angle RCS Evaluation | Sphere | Total Solver Runtime (h)13.5 | 1 | |
| Function Optimization | 20D Sphere Coarse threshold | Convergence Rate0 | 1 | |
| Position Prediction | Sphere Simple | Improvement8.6 | 1 |
Showing 23 of 23 rows