TACO
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Class-level Continual Learning | TACO | AA73.8 | 56 | |
| Code Generation | TACO Verified | During-task Accuracy96.3 | 29 | |
| Fine-grained class-level continual learning | TACO (session=8) | Average Accuracy (AA)71.32 | 11 | |
| Touch Event Detection | TACO 1.0 (test) | mAP48.38 | 8 | |
| Touch Event Detection | TACO (test) | mAP48.38 | 8 | |
| Object Detection | TACO | AR30.3 | 8 | |
| Competitive Programming | TACO (val) | Pass@145.1 | 6 | |
| Dexterous Grasping | TaCo IsaacSim Dexterous Grasping | s_lift (Small)54.7 | 5 | |
| Coding Reasoning | Taco | Pass Rate48.4 | 5 | |
| Exocentric-to-Egocentric Translation | TACO (unseen actions) | FID37.191 | 4 | |
| Code Generation | TACO | Avg@817.9 | 4 | |
| Synthetic Code Data Quality Evaluation | TACO Synthetic Code Data | Originality28.91 | 3 | |
| Hand Pose Estimation | TACO S1 (test) | PA-MPJPE8.02 | 3 | |
| Object Detection | Extended TACO | mAP (0.50:0.95)40.4 | 2 |
Showing 14 of 14 rows