MAZE
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Maze (test)
0.8Success Rate
25
Maze size 8
92.71Success Rate
22
Medium Maze 25x25 world size, 32.8% static obstacle rate
100Success Rate
20
Maze
7.2Path Length Metric 3
19
Maze Hard
97.66Accuracy
18
Maze (test)
98Token F1
16
Maze OOD
70Classification Accuracy
15
Maze ID
93.3Classification Accuracy
15
Maze Hard
93.7pass@1 Accuracy
15
Maze (Standard)
0.996Accuracy
14
Maze
94.4Success Rate
12
Maze Grid Map
258Runtime (ms)
12
Maze Gymnasium
0.97Mean Best Reward
12
Maze
100Score (L=8)
12
10x10 Maze
57Validity Rate
12
Maze size 16
64.85Success Rate
11
Maze size 12
78.31Success Rate
11
Maze size 4
94.38Success Rate (%)
11
Maze (test)
82.1Precision
11
Maze
100RSR
11
Maze
223.55Mean Episode Reward (MER)
11
Maze 2D
90Success Rate
10
Maze 10×10
61.37CR (%)
10
Maze
98Pass@1
10
Maze
96.5Maze Flow (Base)
10