TF+XLA+Mixed Precision: Keras fails to compute gradients

[itmo153277]

Keras fails to compute gradients for autoencoder-esce model using Tensorflow backend with mixed precision and jit compilation enabled.
See code here: colab.

This is caused by UpSampling2D layer. When gradients are computed, the type is resolved as float32 instead of float16, and this causes Relu that comes next to throw a dtype mismatch exception.

The only working workaround I found is explicitly setting dtype to float32 for UpSampling2D layer. This inserts a cast node inbetween relu and upsample which helps dealing with dtype conversion.

Not sure which project this issue should be submitted to: Keras, TF or XLA