A tiny bug during training

[ZhanyuGuo]

Hi, thanks for your great work!

I find a tiny bug which will influence the decay of learning rate: In the function train() in train.py, when epoch reaches freeze_teacher_epoch, it will reset the optimizer and lr_scheduler, which makes epoch reset to 0 in lr_scheduler's view.

I have proved that the lr will never decay in normal training, because step_size=15 and when epoch == 15, lr_scheduler is reset.

I fixed it and trained a new model under the same condition, getting the following results,

	abs_rel	sq_rel	rmse	rmse_log	a1	a2	a3
KITTI_MR	0.098	0.770	4.459	0.176	0.900	0.965	0.983
NEW	0.100	0.755	4.423	0.178	0.899	0.964	0.983

It seems better in sq_rel and rmse.