ValueError: Exception encountered when calling layer "model" (type Functional) about fast-srgan HOT 4 CLOSED

I made a pull request to fix infer.py for tensorflow==2.11.0

Pull Request - #50

Code Changes - 7c2e7c6

Alternatively you can just save the new model and just use it for keras.models.load_model.

    model = keras.models.load_model('models/generator.h5')
    model = change_model(model, new_input_shape=[None, None, None, 3])
    keras.models.save_model(model, 'models/new_generator.h5')

from fast-srgan.

Since I'm not bumping tf version for now, I assume @davlee1972 's fix is enough, closing

from fast-srgan.

Since I'm not bumping tf version for now, I assume @davlee1972 's fix is enough, closing

I think that the code should always be running the latest version of tf. As a user, I think its best if this functionality was integrated into the code base!

from fast-srgan.

@vii empathize with your view, but there are two problems:

1. I currently do not have time to maintain the project
2. TFs versions have seen some drastic breaking changes since the time this was first released.

If you are willing to, I'm happy to accept PRs that bump TF versions while making sure the code doesn't break.

from fast-srgan.