Zero-shot Reinforcement Learning on ExORL RND Walker environment v1 (test)