Open bycn opened 4 years ago
Let's say I want to use a new algorithm, i.e., TD3 HER, and a new replay buffer (i.e., samples goals at saving instead of at training). Which methods would need to be overridden / what's the high level overview of how MPI works for training?
Let's say I want to use a new algorithm, i.e., TD3 HER, and a new replay buffer (i.e., samples goals at saving instead of at training). Which methods would need to be overridden / what's the high level overview of how MPI works for training?