Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Maze Navigation on Maze MCR variant (100 held-out trajectories)
Loading...
0.057
D_policy
PPO
0.03808
0.16579
0.2935
0.42121
Jun 20, 2025
D_policy
Updated 18d ago
Evaluation Results
Method
Method
Links
D_policy
PPO
Reference Policy=Optimal
2025.06
0.057
MPPO B
Reference Policy=Optimal
2025.06
0.076
MPPO A
Reference Policy=Optimal
2025.06
0.084
MPPO A
Reference Policy=Bot A
2025.06
0.471
MPPO B
Reference Policy=Bot B
2025.06
0.481
PPO
Reference Policy=Bot B
2025.06
0.492
PPO
Reference Policy=Bot A
2025.06
0.509
MPPO B
Reference Policy=Bot A
2025.06
0.521
MPPO A
Reference Policy=Bot B
2025.06
0.53
Feedback
Search any
task
Search any
task
