General
BEDD: MineRL BASALT Evaluation Dataset
Recorded human judgments of Minecraft agent videos on four open-ended BASALT tasks.
42,265items
15subjects
MIT, as declared by the Zenodo evaluation-data record and upstream code.license
navigationdomain
planningdomain
videomodality
textmodality
item-level responses released
Response matrix
Loading response matrix…
lowhighUnobserved
Scale: Item-specific grading scales