Implementation question reguarding GlobalMaxPoolGrad

[georgen117]

I have been adding training for the dnnl endpoint. I have been working on Pooling operators but when it came to GlobalMaxPoolGrad I don't see any implementation for GlobalMaxPoolGrad in gradient_builder.cc.

Is GlobalMaxPoolGrad implemented and I am missing it?
If it is not implemented is there a plan to implement it in the future?

[SherlockNoMad]

GlobalMaxPoolGrad is not implemented yet.
You can check the list of training-supported ops in gradient_builder.h

So far, we haven't got any plan to implement a gradient for this op, but we would welcome a contribution for this gradient!