Reinforcement Learning on PROCGEN HEIST 1.0 (test)
98.57Accuracy (%)LAOF-Action
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LAOF-ActionAction Supervision=true2025.11 | 98.57 | 0.91 | |
| LAOM-ActionAction Supervision=true2025.11 | 98.14 | 0.91 | |
| CoMo w/ OFOptical Flow (OF) constraints=true2025.11 | 95.05 | 0.89 | |
| LAOFAction Supervision=false2025.11 | 94.19 | 0.88 | |
| CoMo2025.11 | 75.58 | 0.78 | |
| LAPO2025.11 | 72.23 | 0.76 |