[jira] [Created] (MXNET-1234) Activation gradient (backward pass) has bugs on shape inference

["Pedro Larroy (JIRA)" <ji...@apache.org>]

Pedro Larroy created MXNET-1234:
-----------------------------------

             Summary: Activation gradient (backward pass) has bugs on shape inference
                 Key: MXNET-1234
                 URL: https://issues.apache.org/jira/browse/MXNET-1234
             Project: Apache MXNet
          Issue Type: Bug
          Components: Apache MXNet Backend
            Reporter: Pedro Larroy


Shape inference is not working well for the Activation backward pass and triggering assertions since it has different number of inputs depending on the type of activation such as relu, softsign etc. This logic is not correctly handled and the C++ fail on some build configurations (CPU/MKL/GPU+MKL/GPU etc).



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@mxnet.apache.org
For additional commands, e-mail: issues-help@mxnet.apache.org