model._minimize_model(save_syn1 = True, save_syn1neg = True, save_syn0_lockf = True)
self.model_trimmed_post_training = True
A PR with your proposed incremental fix of not setting up the trimmed flag would be very welcome.
A deprecation warning should also be added. This method will be deprecated in the future as the KeyedVectors is the right way to go for word2vec(not for doc2vec).
Regards
Lev