Real-world data
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| In-context retrieval | Real-world data | SQuAD Accuracy50.1 | 30 | |
| LLM data comprehension | Real-world data | JSON Validity Rate74.3 | 11 | |
| Novel View Synthesis | Real-world Data (test) | PSNR24.64 | 8 | |
| Novel View Synthesis | Real-world data 1.0 (test) | BRISQUE35.94 | 6 | |
| Trajectory Prediction | Real-world data (test) | Success Rate (1m @ 2s)78.04 | 5 | |
| Trajectory Prediction | Real-world data | FDE @ 2s (m)0.76 | 5 | |
| Sparse-view Human Reconstruction | Real-world data | PSNR25.99 | 5 | |
| Pose Estimation | Real-world data Adversarial Conditions | Cube ADD (mm)10.6 | 4 | |
| Pose Estimation | Real-world data Nominal Conditions | Cube ADD (mm)9.1 | 4 | |
| Robotic Object Retrieval | Real-World Data | Accuracy (Simple)54.5 | 4 | |
| Multi-view video generation | Real-world data | FID153.27 | 3 | |
| Joint Auction Mechanism Design | Real-world data Setting E3 | Revenue31.537 | 3 | |
| Joint Auction Mechanism Design | Real-world data Setting E2 | Revenue32.048 | 3 | |
| Joint Auction Mechanism Design | Real-world data Setting E1 | Revenue31.176 | 3 | |
| Action-conditioned video generation | Real-World Data | PSNR38.11 | 2 |
Showing 15 of 15 rows