chongruo opened a new issue #16140: [Feature Request] consistent behavior of 
reshape() in ndarray and symbol mode
URL: https://github.com/apache/incubator-mxnet/issues/16140
 
 
   
   
   For reshape() operator, I found that it works with input which is not a 
tuple. For example:
   
   ```python
   x = mx.nd.array([1,2,3,4])
   
   y = x.reshape(2,2)   # input is not a tuple
   # y = [[1. 2.]
   #     [3. 4.]]
   
   y = x.reshape((2,2))    
   # y = [[1. 2.]
   #     [3. 4.]]
   
   ```
   
   However, the bebavior is not consistent in symbol mode for reshape() with 
non-tuple input. It will raise an error. 
   
   It is bothering that when implementing hybrid_forward() in HybridBlock, 
codes( F.reshape() ) in two modes are the same. We usually disable hybridized 
mode for debugging (printing shape, context), and then enable it for 
efficiency. Since error messages in hybridized mode are a little vague, this 
inconsistent behavior of reshape() makes users hard to realize the reason of 
the error, and also hard to debug. 
   

----------------------------------------------------------------
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