anirudhacharya commented on a change in pull request #13229: Add gauss err
function operator
URL: https://github.com/apache/incubator-mxnet/pull/13229#discussion_r233202588
##########
File path: tests/python/unittest/test_operator.py
##########
@@ -3496,6 +3496,10 @@ def test_special_functions_using_scipy():
mathematical_core("gammaln", lambda x: mx.sym.gammaln(x), lambda x:
scipy_special.gammaln(x),
lambda x: scipy_special.psi(x), 0.5, 0.5)
+ # erf
+ mathematical_core("erf", lambda x: mx.sym.erf(x), lambda x:
scipy_special.erf(x),
Review comment:
yes, but `npout_grad` is calculated from the `backward_numpy_call` method
that we pass to this function.
https://github.com/apache/incubator-mxnet/blob/65736917a63d8d04ee87f653ec2ffcf2be595dd1/tests/python/unittest/test_operator.py#L3473-L3474
with the above statement, the backward pass is getting tested. But that is
not the same thing as doing "gradient checking", right? which will test for the
gradient of the operator without using the `backward_numpy_call`
----------------------------------------------------------------
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
