leezu commented on issue #13100: Hybridization generates superfluous backward ops that may induce storage fallbacks URL: https://github.com/apache/incubator-mxnet/issues/13100#issuecomment-435568020 Also note that adding `F.dot(F.BlockGrad(csr.square()), rsp)` will avoid the storage fallback on CPU. On GPU, adding F.BlockGrad will induce the following error message: ``` [07:42:34] src/engine/engine.cc:55: MXNet start using engine: NaiveEngine Traceback (most recent call last): File "bug.py", line 17, in <module> l = net(csr, rsp) File "/home/ubuntu/mxnet/python/mxnet/gluon/block.py", line 542, in __call__ out = self.forward(*args) File "/home/ubuntu/mxnet/python/mxnet/gluon/block.py", line 909, in forward return self._call_cached_op(x, *args) File "/home/ubuntu/mxnet/python/mxnet/gluon/block.py", line 815, in _call_cached_op out = self._cached_op(*cargs) File "/home/ubuntu/mxnet/python/mxnet/_ctypes/ndarray.py", line 150, in __call__ ctypes.byref(out_stypes))) File "/home/ubuntu/mxnet/python/mxnet/base.py", line 253, in check_call raise MXNetError(py_str(_LIB.MXGetLastError())) mxnet.base.MXNetError: [07:42:38] src/operator/tensor/././cast_storage-inl.cuh:606: Check failed: ctx.requested.size() > 0 (0 vs. 0) ```
---------------------------------------------------------------- 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
