leezu opened a new issue #11552: broadcast_reduce_op does check for negative 
axes
URL: https://github.com/apache/incubator-mxnet/issues/11552
 
 
   
   ## Description
   Consider
   ```python
   import mxnet as mx
   a = mx.nd.ones((1, 1, 1, 1, 1, 1, 1))
   print(mx.nd.broadcast_axes(a, -4, 10).shape)
   ```
   
   Results in
   
   ```
   Traceback (most recent call last):
     File "mxnet_test.py", line 3, in <module>
       print(mx.nd.broadcast_axes(a, -4, 10).shape)
     File "<string>", line 56, in broadcast_axes
     File 
"/Users/lllausen/anaconda3/envs/py36/lib/python3.6/site-packages/mxnet/_ctypes/ndarray.py",
 line 92, in _imperative_invoke
       ctypes.byref(out_stypes)))
     File 
"/Users/lllausen/anaconda3/envs/py36/lib/python3.6/site-packages/mxnet/base.py",
 line 210, in check_call
       raise MXNetError(py_str(_LIB.MXGetLastError()))
   mxnet.base.MXNetError: [00:40:51] 
src/operator/tensor/./broadcast_reduce_op.h:306: Check failed: 
oshape[param.axis[i]] == 1U (2332991486559351923 vs. 1) Broadcasting axis must 
have size 1
   
   Stack trace returned 8 entries:
   [bt] (0) 0   libmxnet.so                         0x000000011000c3b4 
libmxnet.so + 21428
   [bt] (1) 1   libmxnet.so                         0x000000011000c16f 
libmxnet.so + 20847
   [bt] (2) 2   libmxnet.so                         0x0000000110359431 
libmxnet.so + 3482673
   [bt] (3) 3   libmxnet.so                         0x000000011113d47a 
MXNDListFree + 497978
   [bt] (4) 4   libmxnet.so                         0x000000011113c099 
MXNDListFree + 492889
   [bt] (5) 5   libmxnet.so                         0x000000011109d40a 
MXCustomFunctionRecord + 20666
   [bt] (6) 6   libmxnet.so                         0x000000011109e4b0 
MXImperativeInvokeEx + 176
   [bt] (7) 7   libffi.6.dylib                      0x000000010f0f7884 
ffi_call_unix64 + 76
   ```
   
   The expected behavior is to either check and report that negative axes are 
invalid or to wrap around.

----------------------------------------------------------------
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:
[email protected]


With regards,
Apache Git Services

Reply via email to