Blocksworld Dataset

The Blocksworld dataset is a photo-realistic environment, where the goal is to move blocks around to achieve a goal state. The dataset consists of 480/2592 possible states/transitions, which are used to train the First-Order State AutoEncoder.

Data and Resources

Cite this as

Masataro Asai, IBM Research, MIT-IBM Watson AI Lab (2024). Dataset: Blocksworld Dataset. https://doi.org/10.57702/olg828xm

DOI retrieved: December 2, 2024

Additional Info

Field Value
Created December 2, 2024
Last update December 2, 2024
Defined In https://doi.org/10.48550/arXiv.1902.08093
Author Masataro Asai
More Authors
IBM Research
MIT-IBM Watson AI Lab