You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Indeed, state_part_zoneout_prob is referenced in call but isn't initialized anywhere in the float version.
However, if you happen to have your zoneout probabilities in a numpy array and the state_size is a tuple, the computation will actually go through without problems. The code should still be fixed, but at least there is a working path.
As I missing something, or both zoneout_tensorflow and zoneout_seq2seq feature non-existing variables and wrong function parameter names?
The text was updated successfully, but these errors were encountered: