kshitij12345 commented on a change in pull request #14992: [MXNET-978] Support
higher order gradient for `log`.
URL: https://github.com/apache/incubator-mxnet/pull/14992#discussion_r285357016
##########
File path: src/operator/tensor/elemwise_unary_op_basic.cc
##########
@@ -1016,7 +1016,28 @@ The storage type of ``log2`` output is always dense
.set_attr<nnvm::FGradient>("FGradient", ElemwiseGradUseIn{"_backward_log2"});
MXNET_OPERATOR_REGISTER_BINARY_WITH_SPARSE_CPU_DR(_backward_log,
-
unary_bwd<mshadow_op::log_grad>);
+
unary_bwd<mshadow_op::log_grad>)
Review comment:
I don't know much about this library but,
I believe it would be better to have gradients defined for existing
backward, instead of a differentiable gradient (relying on autograd machinery)
at least on `ops` where backward is not trivial. It will allow to use existing
optimised fused kernels and make sure there is no regression in the backward.
Note: `log` is relatively trivial (single reciprocal). But maybe we may see
a performance regression for `sigmoid`, if we do it by relying on autograd
machinery instead of the existing `_backward_sigmoid`.
----------------------------------------------------------------
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
