Add to Favourites
To login click here
Powderworld is a simulation environment developed by two researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) to support agent learning and multi-task generalization. It runs directly on the GPU and includes two frameworks for specifying world-modeling and reinforcement learning tasks. Powderworld is beneficial to the RL community as it provides a “foundation environment” that allows a variety of tasks originating from the same core rules, making it easier to compare different training task variations.
