Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

problem with module bind #1

Closed
xiuliren opened this issue Aug 21, 2018 · 3 comments
Closed

problem with module bind #1

xiuliren opened this issue Aug 21, 2018 · 3 comments

Comments

@xiuliren
Copy link

xiuliren commented Aug 21, 2018

Hi, thanks for making this package.
I just tried, but have some problems. I am not familiar with mxnet, but would like to try. Maybe there is a simple fix.

I have compiled the master branch of mxnet using cuda 9.2 and cudnn 7.

Traceback (most recent call last):                                                                    
  File "pointcnn_cls.py", line 136, in <module>                                                       
    mod.bind(data_shapes=[('data',(batch_size_train, sample_num_train,3))], label_shapes=[('softmax_la
bel',(batch_size_train, probs_shape[1]))], for_training=True, shared_module=mod)                      
  File "/opt/incubator-mxnet/python/mxnet/module/module.py", line 416, in bind                        
    shared_module.binded and shared_module.params_initialized                                         
AssertionError                                                                                        

If I change the shared_module=mod to shared_module=None, the program runs but it seems that it is not learning anything. I have ran for about a hour, and the value keep about 0.1 which is random guess.

@chinakook
Copy link
Owner

chinakook commented Aug 22, 2018

It's caused by the recent PR: apache/mxnet#12155
I'll fix it later.

@chinakook
Copy link
Owner

It's fixed in my master branch. @jingpengw

@xiuliren
Copy link
Author

thanks a lot! It works great!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants