| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Action-conditioned World Modeling | EgoDex (test) | PSNR26.78 | 10 | |
| Image-to-Video generation | EgoDex Eval | PSNR23.69 | 9 | |
| Trajectory Prediction | EgoDex | ADE0.061 | 9 | |
| Action-conditioned world modeling | EgoDex 50 (test) | PSNR25.74 | 8 | |
| Value-Order Correlation | EgoDex expert demonstrations | VOC95 | 7 | |
| 3D Hand-Controlled Video Generation | EgoDex 1.0 (test) | FID39.62 | 7 | |
| Video Frame Rank-Correlation | EgoDex | VOC Score (Sparse)0.88 | 6 | |
| Text-conditioned World Modeling | EgoDex (test) | PSNR21.05 | 5 | |
| Egocentric video simulation | EgoDex Tabletop Scene (test) | PSNR25.056 | 5 | |
| Temporal Value Estimation | EgoDex (test) | VOC+80.67 | 5 | |
| Long-horizon Video Prediction | EgoDex long-horizon 65 frames | PSNR22.75 | 4 | |
| Video Generation | EgoDex | PSNR19.84 | 4 | |
| Action-conditioned Video Generation | EgoDex (test) | FVD39.58 | 4 | |
| Action-conditioned egocentric video generation | EgoDex | PSNR19.35 | 4 | |
| Dexterous manipulation | EgoDex-R | Object Rotation Error (Er)9.7 | 3 | |
| Trajectory Forecasting | EgoDex | ADE@40.071 | 3 | |
| Open-Loop Trajectory Prediction | EgoDex (test) | Embedding L2 Error (At 4s)0.62 | 3 | |
| Text-to-Video-Action Generation | EgoDex | FVD191.1 | 2 | |
| Controllable Video Generation | EgoDex | PSNR28.06 | 2 |