mseeger commented on issue #8323: clean up math operators
URL: https://github.com/apache/incubator-mxnet/pull/8323#issuecomment-338201650
Hello, if you don't mind, I take this over, and add something on top.
I'll return to work from my leave next Wed, and would like to have all of
this
mseeger commented on issue #8323: clean up math operators
URL: https://github.com/apache/incubator-mxnet/pull/8323#issuecomment-337948535
Hi Eric, I did some test to confirm that this solution works.
Please do merge it in. Otherwise, if you are too busy, I can take it over.
mseeger commented on issue #8323: clean up math operators
URL: https://github.com/apache/incubator-mxnet/pull/8323#issuecomment-337679717
Well if this works, it is fine. Just make sure that if the arg is double,
then the computation is done in double. But I think my new unit tests would
mseeger commented on issue #8323: clean up math operators
URL: https://github.com/apache/incubator-mxnet/pull/8323#issuecomment-337660985
Independent of this PR, I'd like to do one more change.
I am not happy with many of the backward expressions: different to the
forward ones, they
mseeger commented on issue #8323: clean up math operators
URL: https://github.com/apache/incubator-mxnet/pull/8323#issuecomment-337591796
I am looking at pages like this:
https://stackoverflow.com/questions/15911890/overriding-return-type-in-function-template-specialization
mseeger commented on issue #8323: clean up math operators
URL: https://github.com/apache/incubator-mxnet/pull/8323#issuecomment-337589379
It is funny this should work. If you define float name(DType a) in the
templated class, I thought you cannot specialize this to be double name(double
