search

ponder-lab / Hybridize-Functions-Refactoring

Refactorings for optimizing imperative TensorFlow clients for greater efficiency.
Eclipse Public License 2.0
0 stars 0 forks source link

Can't infer Python side-effects for tf.keras.Models with implicit callables #291

Open khatchad opened 11 months ago

khatchad commented 11 months ago

Consider the following model:

https://github.com/ponder-lab/Hybridize-Functions-Refactoring/blob/6b7717c696a090326fa200b827b8b496782d7faf/edu.cuny.hunter.hybridize.tests/resources/HybridizeFunction/testModel2/in/A.py#L6-L30

And the following client code:

https://github.com/ponder-lab/Hybridize-Functions-Refactoring/blob/6b7717c696a090326fa200b827b8b496782d7faf/edu.cuny.hunter.hybridize.tests/resources/HybridizeFunction/testModel2/in/A.py#L39-L40

This code implicitly calls SequentialModel.call() because Model's __call__() method calls Model.call() and SequentialModel overrides call().

khatchad commented 11 months ago

Blocked on https://github.com/wala/ML/issues/106.