Continuous Control on MAZE
94.4Success RateLC-DP
Evaluation Results
| Method | Links | |
|---|---|---|
| LC-DPFamily=Language-critique (DP), Supervision=Language labels2026.07 | 94.4 | |
| EDPFamily=Diffusion-based, Supervision=Scalar rewards2026.07 | 91.2 | |
| LC-BCFamily=Language-critique (BC), Supervision=Language labels2026.07 | 81.6 | |
| TD3+BCFamily=Policy-regularized, Supervision=Scalar rewards2026.07 | 81.2 | |
| DTFamily=Sequence-modeling, Supervision=Scalar rewards2026.07 | 41.6 | |
| CQLFamily=Value-based, Supervision=Scalar rewards2026.07 | 30 |