Closed 1tac11 closed 1 year ago
self.rnn_states is the current rnn state before next step. mb_rnn_states are saved for training rnn states. For example we have 1 env with horizon_length 64 and seq_length 16. It means we will split 64 into 4 sequences with length 16 and we need to have some start rnn state - mb_rnn_states.
hi there, what does the value
self.mb_rnn_states
stands for?self.rnn_states
is already set, but this value is updated when horizon episode is finished.kind regards