ScienceWorld
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
ScienceWorld
32.43Mean Normalised Score
18
ScienceWorld
75L0 Score
18
ScienceWorld (test)
38.18Success Rate
17
ScienceWorld (val)
44.08Success Rate
17
ScienceWorld
29.5Score
16
ScienceWorld
11.5Efficiency Factor
15
ScienceWorld Unseen
58.5Average Reward
15
ScienceWorld (OOD)
9.9Score
14
ScienceWorld (test)
53.8Score
14
ScienceWorld
67.2L0 Score
13
ScienceWorld
90Average Reward
12
ScienceWorld
41.07Success Rate
11
ScienceWorld Continual-memory deployment
41.1Success Rate
11
ScienceWorld
67.5Success Rate
11
ScienceWorld Seen
71.6Average Reward
10
ScienceWorld Unseen
75.67Average Reward
9
ScienceWorld Seen
77.1Average Reward
9
ScienceWorld OOD (test)
57.81Success
9
ScienceWorld
50.34Accuracy
8
ScienceWorld
96.06Base Score
8
ScienceWorld v1.0 (test)
97.04Task 1-1 (L) Score
8
ScienceWorld standard 30-task protocol
73.72Average Score (Short)
7
ScienceWorld (Unseen)
0.586Success Rate
7
ScienceWorld (test)
44.8Task 1-1 Score
7
ScienceWorld variations (test)
4Changes of State: Boiling Success
7