ea98b39996
This is something tf.keras will do. I'm not totally sure about dtype, we could only save it if it diverges from the global policy. But when it is explicitly set on the layer, it is probably important to persist. |
||
---|---|---|
.. | ||
__init__.py | ||
activity_regularization_test.py | ||
activity_regularization.py | ||
dropout_test.py | ||
dropout.py | ||
gaussian_dropout_test.py | ||
gaussian_dropout.py | ||
gaussian_noise_test.py | ||
gaussian_noise.py | ||
spatial_dropout_test.py | ||
spatial_dropout.py |