You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@mxnet.apache.org by GitBox <gi...@apache.org> on 2018/10/29 17:57:58 UTC

[GitHub] leleamol commented on issue #12966: Strange behaviour/possible bug with mxnet::cpp::Symbol::Variable name argument

leleamol commented on issue #12966: Strange behaviour/possible bug with mxnet::cpp::Symbol::Variable name argument
URL: https://github.com/apache/incubator-mxnet/issues/12966#issuecomment-434014611
 
 
   @nikudyshko, @ddavydenko  I spent some time experimenting with this issue. This is an interesting issue.  If I use 2 or 3 characters in following lines to name the variable (e.g ws or wst, bs or bst) then the training fails.
   weights[i] = Symbol::Variable("ws" + std::to_string(i));
   biases[i] = Symbol::Variable("bs" + std::to_string(i));
   
   However, if I use 1 or more than 3 characters then the training proceeds correctly.
   I looked at implementations c++ package API implementations briefly and nothing stood out.  I would have to dive deeper.  Would also need to see if it is reproducible in python.
   
   Will provide an update. 
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services