Model not learning with Tensoflow2

[Kamnitsask]

After communications with a user (thanks Shadab), we found that if this code is run using Tensorflow-2, training the model does not lead to learning/convergence. In fact, error goes super-high, eg 90% on mnist.

On his same settings, simply downgrading to Tensorflow 1.13 (where the code was written) solves this and everything works like normal.

Please use Tensorflow-1 for now (I ve tested till 1.13).

I will try to make time and update my projects to TF2 in the next few months (hopefully Dec'19 + Jan'20). But maintaining all projects takes quite some time so please excuse delays.

Thank you,
Konstantinos

[wanyu-lin]

Dear author, your work is pretty interesting. I am wondering if it works with TF2 now?

[Kamnitsask]

Hello, glad you find the method interesting :)

The code has not been updated to be compatible with TF2. To use it, I would suggest you simply create a virtual environment with a TF1.XX version.

Sorry for that, I will try to keep in my mind to update the code, but the workaround seems relatively easy so I ve kept this task relatively low priority.