MolGan tutorial gives NAN

[mghahremanpour]

Hi,

I am running the MolGan model provided at https://github.com/keras-team/keras-io/blob/master/examples/generative/molecule_generation.py, but training gives NaN when more than one convolution layer is used in the encoder. I was wondering if you could help me with this.

Thanks

[sachinprasadhs]

Hi, If you are modifying the existing code, could you please provide the code Gist to replicate the reported behavior. Thanks

[github-actions]

This issue is stale because it has been open for 14 days with no activity. It will be closed if no further activity occurs. Thank you.

[mghahremanpour]

Code is not modified!

[github-actions]

This issue is stale because it has been open for 14 days with no activity. It will be closed if no further activity occurs. Thank you.

[github-actions]

This issue was closed because it has been inactive for 28 days. Please reopen if you'd like to work on this further.

[github-actions]

Are you satisfied with the resolution of your issue?
Yes
No

[sachinprasadhs]

@mghahremanpour , Could you point us to the part where you are seeing NaN

[mghahremanpour]

The loss value becomes NaN with more than 2 conv layers in get_encoder