Why use the" y_eval" of the tuning parameter as the test set for evaluation, shouldn't it be separate?

[maymodric]

[Ty4Code]

I totally agree, it is likely not a good idea in general to use the test set as our 'validation_data' input for the TensorFlow fit function.

However, in this particular case, I don't think it's an issue because there is no early stopping and the validation loss is not used for any hyperparameter tuning.

In other words, if we were to remove the validation_data=(x_eval, y_eval) line from the notebook, it would not have any impact on the final results of the training run/notebook run as far as I can tell.