This code performs SSX algorithm on Minipacman. 

This code is based on a github repository available at:
https://github.com/higgsfield/Imagination-Augmented-Agents. All files in common are directly from the Imagination-Augmented-Agents github repository with some modifications for SSX.

Command to run SSX on minipacman:

python minipacman_savefigs.py --mode mode --scenario scenario

where mode is either eat or hunt and scenario is 1, 2, or 3. Directories will be created in figures/strategic_states with the results of policy eat or hunt applied to scenario 1, 2, or 3. SSX will be applied to several states along the path of the saved scenario. Saved scenarios are in directory saved_boards. Pretrained eat and hunt models are in directory models.