Textworld
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
TextWorld
100Real
24
TextWorld
74.28Mean Normalised Score
18
TextWorld (TW)
70.6EM Accuracy
16
TextWorld Cooking (test)
75.5Accuracy
14
TextWorld Treasure (test)
81.5Accuracy
14
TextWorld Quest (test)
88Accuracy
14
TextWorld
100Real
14
Textworld
75Success Rate
12
TextWorld
98.7Success Rate (%)
6
TextWorld Cooking
76Accuracy
6
TextWorld Treasure
81Accuracy
6
TextWorld Quest
88Accuracy
6
TextWorld Cooking
0.78Mean Episodic Return
5
TextWorld Cooking (test)
—Primary metric
0