ScienceWorld
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
ScienceWorld (ETO)
0.97Rho
6
ScienceWorld Llama-3.1 8B backbone (Out-of-Distribution (OOD))
34.36Performance
6
ScienceWorld Llama-3.1 8B backbone (In-Distribution (ID))
67.61Performance
6
ScienceWorld Unseen
10.6Average Steps
6
ScienceWorld Seen
10.2Average Steps
6
ScienceWorld
88Running Max Return
6
ScienceWorld 30 tasks
84.7Score
6
ScienceWorld
35Accuracy
5
ScienceWorld (Seen)
0.85Average Reward
5
ScienceWorld (test)
63.35Score
5
ScienceWorld
44.67Mean Score (Task 10-2 (L))
5
ScienceWorld
100Mean Score
5
ScienceWorld
72.33Mean Score
5
ScienceWorld
56.67Mean Score
5
ScienceWorld
60Mean Score (Task 9-2 (L))
5
ScienceWorld
76.67Mean Score
5
ScienceWorld
64.44Mean Score (ScienceWorld Physics)
5
ScienceWorld
20Mean Score (Task 8-2 (S))
5
ScienceWorld
38Mean Score (Task 8-1 S)
5
ScienceWorld
83Mean Score
5
ScienceWorld
83.33Mean Score (Task 7-2)
5
ScienceWorld
100Mean Success Rate (Task 7-1 (S))
5
ScienceWorld
57.8Mean Score
5
ScienceWorld
77Mean Score
5
ScienceWorld
83.33Mean Score
5