Grzegorz Dudziuk

Than you very much for doing this. Let me post also the slightly modified test code with built-in memory measurements, which may be more convenient: ```python import tensorflow as tf...

Also, anybody investigating the root of the problem, be sure to note the issue tensorflow/tensorflow#35030, where @mihaimaruseac has tracked the point at which the bug has been introduced.

Not stalled. I will try the custom training loop next week.

I have checked that the issue is still there in TF 2.8 with Python 3.8.

Now, let me answer @rchao's questions. First, yes, this is a regression from previous versions. As stated in the initial post of this issue, the issue first occurred in `2.0.0-rc0`....

Second, the custom training loop. I have never used custom training loops in TF before but it turned out to be quite straightforward by modifying the examples from official tutorials....

The code with the custom training loop: ```python import tensorflow as tf import numpy as np import psutil import os from tensorflow.keras.models import Model from tensorflow.keras.layers import Input, Lambda, Conv2D...

And the output is below (TF 2.8, Python 3.8). The memory usage is slightly lower than with `model.fit` (compare to the results posted in tensorflow/tensorflow#40942) but still ca. 2x the...

How is that possible that this issue has been qualified as stalled? I have answered @rchao 's questions and was waiting for response.

An update. As long as this implementation relies on the form `(Eq. 2)` of the target function, it is important that `{std}_j` do not depend on the dynamic parameters `r`...