Skip to content

training readout does not work #19

Description

@jackklpan

Hi,

I meet two problems when training the readout network.

return self._do_restrict_information(x, alpha)

does not work now, since
def _do_restrict_information(self, x):

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

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!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions