some question about fuction get_refine_net in molopt_score_model.py

[zhengchen0914]

when i study you code, i find the fuction get_refine_net is configured in the config file as “uni_o2“, which seems to mean that the egnn module is not used in training, which is inconsistent with the description in the paper.
I'm not sure if there is something wrong with my understanding, and I hope someone can answer my questions.
thanks