I am using Tensorflow 2.15.1 and, if I use model.save_weights("./dir/mymodel.bin") and subsequently try to load the weights with model.load_weights (in a different process), the model does not return the expected results (when using model.evaluate or .predict). My model has a custom loss function and metric, but this should not be a problem, since I try to load weights only. Any idea why I get this error?
Comment From: sachinprasadhs
Could you please try with the Keras 3 version and let us know your observations and possibly some reproducible code. Thanks
Comment From: gcelano
I installed Tensorflow 2.16.1, but the it cannot detect the GPUs.
Comment From: gcelano
The problem is also with Tensorflow 2.16. Please, run the code here:
https://git.informatik.uni-leipzig.de/celano/dithrax_test
Comment From: sonali-kumari1
Hi @gcelano -
Could you please provide a standalone code in the form of a colab notebook to reproduce this issue? Thanks!
Comment From: github-actions[bot]
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.