training readout does not work

[jackklpan]

Hi,

I meet two problems when training the readout network.

IBA/IBA/pytorch_readout.py

Line 128 in 34baed6

return self._do_restrict_information(x, alpha)

does not work now, since

IBA/IBA/pytorch.py

Line 412 in 34baed6

def _do_restrict_information(self, x):

accepts one argument now.
(Currently, I checkout to the previous commit)

IBA/IBA/pytorch_readout.py

Line 170 in 34baed6

alpha.clamp(-self._alpha_bound, self._alpha_bound)

should be alpha = alpha.clamp(....
(The alpha become -infinite)

Could you check the current code can train the readout network?
Thanks!

[karl-schulz]

Hi @jackklpan !

You are right, there were some issues I hope to have resolved with the latest pull request #21 . We will add a test for the readout IBA so this doesn't happen again.

Regarding the stability of the training, good spot! We also improved a minimal std in the normalization step to prevent divsion by very small numbers.

Please check if this solves your issues, we will do the same. We are happy to hear about your results.

Thanks for your contribution!

[jackklpan]

Thanks! I will try to train again.

[voa18105]

Hello! Actually, there is a same issue with "per sample" - no alpha given