farazsaeedan

Hi. which variant of DPP have you used? and how many instances of DPP did you place in ResNet-101?

I do not have any implementation for Tensorflow but for PyTorch I already have a folder in place and will try to populate it with "mainscript" PyTorch and also some...

dividing x by xn is just for numerical stability and does not change the value of the output result as it cancels out. for the DPP complete implementation look at...

normalizing "x"s is done for increased numerical stability.

This looks like a problem with nn (or cunn) libraries. I suggest trying to reinstall a "clean" version of these libraries as suggested here: [link1](https://github.com/facebookresearch/deepmask/issues/5) or here [link2](https://github.com/torch/cunn/issues/471)