FrozenLake
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
FrozenLake reward reversal Hidden drift
100Score
45
FrozenLake (Source)
67.5Score
36
FrozenLake Source
9,000Success Rate
27
FrozenLake
100Success Rate (Level 3)
19
FrozenLake Drift II
8,500Success Rate
18
FrozenLake v1.0 (Drift I)
7,500Success Rate
18
FrozenLake Drift II - Topology Shift
7,500Success Rate
18
FrozenLake Drift I - Topology Shift
8,500Success Rate
18
FrozenLake reward reversal Source
90Score
18
FrozenLake
82Success Rate
17
FrozenLake
100Success Rate
17
FrozenLake (FL) (test)
0Loop Ratio
17
FrozenLake OOD
90Classification Accuracy
15
FrozenLake ID
100Classification Accuracy
15
FrozenLake
83Success Rate
15
FrozenLake Implicit Hidden Drift
88.8Success Rate (Source)
14
FrozenLake Explicit Structural Drift II
83.7Success Rate (Source)
14
FrozenLake
27Success Rate
12
FrozenLake
0.94Reward
12
FrozenLake OOD
98Success Rate
11
FrozenLake size 16
49.38Success Rate
11
FrozenLake size 12
64.72Success Rate
11
FrozenLake size 8
81.74Success Rate
11
FrozenLake size 4
87.94Success Rate
11
FrozenLake held-out (val)
60.7Success Rate
10