What is the difference between TensorFlow's callbacks and early stopping?
Early stopping and callbacks are two different concepts:
Early stopping is a machine learning concept about when to stop training your model to avoid overfitting: You monitor a target value (e.g. validation loss) and stop learning after it hits a minimum. If the monitored value keeps increasing for a couple of epoch, you can restore the weights from the minimum.
Callbacks are a general software engineering concept and a technical implementation detail on how to trigger certain functions.
A callback function (e.g. early stopping) is a function that is passed as an argument to another function (e.g. fitting an ML model), and called in certain situations (e.g. after finishing an epoch). The callback function will perform its business (e.g. monitor the validation loss and decide whether to trigger the end of the training) and return control to the calling function (fitting the model).
In TensorFlow, early stopping is implemented using a callback function, but this is not the only way to do it. Also, there are further features in TensorFlow that are implemented using callbacks.