wgpubs
(WG)
December 16, 2018, 7:12pm
1
I built a simple callback that outputs the ROC/AUC score at the end of training via print(...)
, but it now overwrites the output generated by learn.fit_one_cycle()
.
Is there a way I can just append this output to that displayed as a result of calling fit()
?
1 Like
wgpubs
(WG)
August 28, 2019, 7:45pm
2
Iām not sure why, but appears that my call to self.learn.get_preds()
is the culprit:
def on_epoch_end(self, **kwargs:Any)->None:
preds, targets = self.learn.get_preds()
print('test')
If I remove the call to get_preds
, the printed text is appended to the output rather than replacing everything.
Any ideas how to fix?