def retrain(self, batch_size): minibatch =random.sample(self.experience_replay, batch_size) for state, action, reward, next_state in minibatch: state = np.asmatrix ...

At this point I've tried to reproduce it on different desktop but ran into same issue. I don't know what could be the reason. I've also tried to download model and load it via BertEmbeddings.load() ...