yulequan / UA-MT

code for MICCAI 2019 paper 'Uncertainty-aware Self-ensembling Model for Semi-supervised 3D Left Atrium Segmentation'.
https://arxiv.org/abs/1907.07034
479 stars 97 forks source link

Why is the student model stored instead of a teacher model? #8

Closed chuxiang93 closed 4 years ago

chuxiang93 commented 4 years ago

Dear @yulequan Another question, why is the student model stored instead of a teacher model in the training phase. In the paper of "Mean Teacher", I found that the teacher model has better performance. if iter_num % 1000 == 0: save_mode_path = os.path.join(snapshot_path, 'iter_' + str(iter_num) + '.pth') torch.save(model.state_dict(), save_mode_path) logging.info("save model to {}".format(save_mode_path)) Thanks, Jianqiang Ma

yulequan commented 4 years ago

Hi @mjq93, In our experiment. The student model and teacher model achieve similar performance. Therefore, we both use student model for final evaluation.

chuxiang93 commented 4 years ago

OK, I know, thank you for your reply.