Frozen Lake
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Frozen Lake 8x8
3.53Cumulative Reward
20
Frozen Lake 4x4 Hard
8.95Reward Value
20
Frozen Lake 4x4 Default
62.42Total Reward
20
Frozen Lake
69Pass@1 Success Rate
10
Frozen Lake
98.7Success Rate (pass@2, 4x4)
10
Frozen Lake
1.985ATE
9
Frozen-Lake 8x8 Grid
12.5Accuracy
9
Frozen-Lake 6x6 Grid
25Accuracy
9
Frozen-Lake 4x4 Grid
55Accuracy
9
Frozen Lake Random 1
0.37Max Score
8
Frozen Lake
95.9Success Rate
7
Frozen Lake standard_4x4
1Total Reward
4
Frozen Lake standard_4x4
100Critical State Safety Rate
4
Frozen Lake (test)
38.3Accuracy
4
Frozen Lake
0.398Mean Execution Time (s)
2
Frozen Lake Catastrophic Path
-87Lowest Cumulative Reward
2
Frozen Lake Almost Catastrophic
-68Lowest Cumulative Reward
2
Frozen Lake Slightly Suboptimal Path
41Lowest Cumulative Reward
2
Frozen Lake
37.3Average Worst-Case V(s0)
2
Frozen Lake 128 samples (test)
6.3Accuracy
1