zixuanweeei opened a new pull request #17872: Fix issue of zeros gradients 
w.r.t. RNN bias when num_layers > 1
URL: https://github.com/apache/incubator-mxnet/pull/17872
 
 
   ## Description ##
   Patch for the issue #17818. The rnn operator produces zero gradients for 
bias when num_layers > 1. It is caused by a mistake in calculating the shift of 
bias pointer, where we used the size of fusion bias (i2h_bias + h2h_bias) but 
MXNet gives twice (i2h_bias, h2h_bias) as many as the fusion size.
   
   ## Checklist ##
   ### Changes ###
   - [x] Use the correct shift of bias pointer.
   - [x] Change the way of sharing the same values of parameters of the fused 
RNN layers and the stacked one.
   - [x] Add check for RNN output states.
   
   @ciyongch @pengzhao-intel @TaoLv 
   

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


With regards,
Apache Git Services

Reply via email to