Tensorflow model.fit fails on test_step: 'NoneType' object has no attribute 'items'

[JVD9kh96]

I am using tf.data module to load my datasets. Although the training and validation data modules are almost the same. The train_step works properly and the training on the first epoch continues till the last batch, but in the test_step I get the following error:

    353     val_logs = {
--> 354         "val_" + name: val for name, val in val_logs.items()
    355     }
    356     epoch_logs.update(val_logs)
    358 callbacks.on_epoch_end(epoch, epoch_logs)

AttributeError: 'NoneType' object has no attribute 'items'

Here is the code for fitting the model:

results = auto_encoder.fit(
    train_data,
    epochs=config['epochs'],
    steps_per_epoch=(num_train // config['batch_size']),
    validation_data=valid_data,
    validation_steps=(num_valid // config['batch_size'])-1,
    callbacks=callbacks
)

I should mention that I have used .repeat() on both train_data and valid_data, so the problem is not with not having enough samples.

[mehtamansi29]

Hi @JVD9kh96 -

Can you help me with full code with dummy dataset to reproduce error ?

[google-ml-butler]

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