How to render zero-shot transfer policy? about hierarchical_morphology_transfer HOT 2 CLOSED

Once you have trained a high level policy, it will additionally store the low level policy. You just need to run the test script on the high level policy. One note is that you shouldn't train the low level for Ant on AntMaze_Low, as this will train the Ant with the maze walls in place. Rather, it's best to train the ant low policy on just the Ant_Low environment.

from hierarchical_morphology_transfer.

@jhejna Thanks for you reply! Okay, all I need to do is to train a PointMaze_High and Ant_Low, then combine them with composition_test.py script.

from hierarchical_morphology_transfer.