Could you please help me answer this? I'm currently trying to replicate the experiments in the paper.
I would like to inquire about the experimental scores in the paper. For instance, for the game "Sokoban", was it run using the "harness: true" , "observation_mode: vision" settings ?
