Rumor: disproved!
tf.keras *is* our recommended higher-level API for @TensorFlow 2.0, but tf.estimator will still be around for the life of 2.0.
If you've got estimators you feel strongly about, you'll still be able to use 'em. 
https://twitter.com/mathtick/status/1113112239765118978 …
-
Thanks! Good to know. I think it is the single batch dim of Keras that is keeping people in tf.estimator land, possibly some tensorboard interactions too.
I'm not sure what you mean, but I'd like to figure it out. What is "single batch dim"?
-
-
Just mean multi-inputs where they might all have different dimensions. For example, you get a batch of samples S at some fixed grid of locations x. You give the model [S, x]. Stackover-lore seems to indicate this kind of thing fights Keras but maybe adding more dim is the trick?
-
Oh and I think the response here is relevant, I'll try to follow up there with any notes that come out of this https://stackoverflow.com/questions/44101692/implementing-a-multi-input-model-in-keras-each-with-a-different-sample-sizes-ea …
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.