Nice work for this benchmark, and I am working on this transfering this benchmark to my custom environment. I want to enquire that how can I identify that the unsupervised RL algorithm truely work rather than some random trajectories? Any metrics that can help me identify that?
Nice work for this benchmark, and I am working on this transfering this benchmark to my custom environment. I want to enquire that how can I identify that the unsupervised RL algorithm truely work rather than some random trajectories? Any metrics that can help me identify that?