Reputation: 1854
Is there a way to pass options
and run_metadata
to tf.Session().run()
while still using Keras.Model.fit()
?
The only way I see to use those run options is to take control over the TF session from Keras by implementing the training loop myself. But I was I wondering if you know of a way that avoids that.
Upvotes: 1
Views: 1618
Reputation: 5206
Model.compile
takes keyword arguments which are passed to session.run
.
Upvotes: 2