You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.
For BN layers, this will erase the moving_mean and moving_var in aux, which will lead to wrong results; I don't know if there are any special reasons to do this? @precedenceguo
The text was updated successfully, but these errors were encountered:
zhangjiangqige
changed the title
a question about rcnn exmaple, why setting aux to zero when testing rpn?
a question about rcnn exmaple, why setting aux to zero when testing?
Aug 9, 2016
There is not. Please ignore this line and supply aux_params as saved in training.
zhangjiangqige
changed the title
a question about rcnn exmaple, why setting aux to zero when testing?
a question about rcnn exmaple, why set aux to zero when testing?
Aug 9, 2016
https://github.com/dmlc/mxnet/blob/master/example/rcnn/rcnn/rpn/generate.py#L30
For BN layers, this will erase the moving_mean and moving_var in aux, which will lead to wrong results; I don't know if there are any special reasons to do this? @precedenceguo
The text was updated successfully, but these errors were encountered: