Counterfactual Policy Evaluation on GridWorld p = 0.4
230Worst-Case Counterfactual V(s0)Gumbel-max
Evaluation Results
| Method | Links | |
|---|---|---|
| Gumbel-maxSampled observed trajectories=1002025.02 | 230 | |
| ICFMDPSampled observed trajectories=1002025.02 | 81.4 |
| Method | Links | |
|---|---|---|
| Gumbel-maxSampled observed trajectories=1002025.02 | 230 | |
| ICFMDPSampled observed trajectories=1002025.02 | 81.4 |