This repository has been archived by the owner on Nov 17, 2023. It is now read-only.
Replies: 2 comments
-
@mxnet-label-bot add [documentation, question] |
Beta Was this translation helpful? Give feedback.
0 replies
-
@mxnet-label-bot add [doc] |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
The documentation of
mx.sym.MakeLoss
is highly confusing. To my understanding the only thingMakeLoss
does is that it wraps an existing symbol and defines that it does not require a head gradient when used in optimization. Further, it seems that the output of theforward()
call to a MakeLoss-symbol is the output of theforward()
call of the wrapped symbol. That is, MakeLoss just passes through its input data inforward()
.However, the documentation states the following:
"The output of this function is the gradient of loss with respect to the input data."
What does this mean? I read it as if the output of
forward()
is the same as the output ofbackward()
, namely the gradient of the symbol, MakeLoss wraps. But this does not seem to be true.Beta Was this translation helpful? Give feedback.
All reactions