Understanding filepath naming

Question

Just looking at the documentation of ModelCheckpoint

I get that ".2f" probably means "take the value up to the second decimal digit".

Where can I find a complete reference of the syntax for my model saves?

That's just Python's normal `str.format()` options, not anything specific to Keras. — jasonharper, Jun 10 '21 at 14:45
Oh. Ok, thanks, I didn't know that syntax. If you want to answer it, I'll accept it. — Tartempion, Jun 10 '21 at 14:58
So I assume I can only use parameters of the model in the brackets and nothing from outside? — Tartempion, Jun 10 '21 at 14:59

score 0 · Answer 1 · answered Jun 10 '21 at 15:28

@jasonharper already mentioned in his comment that these are simply the options for Python's string formatting. To elaborate on the options you have, let's first check the documentation for the filepath argument:

filepath can contain named formatting options, which will be filled the value of epoch and keys in logs (passed in on_epoch_end).

This directs us to the on_epoch_end method of TensorFlow Callbacks which has the following arguments:

epoch: Integer, index of epoch.

logs: Dict, metric results for this training epoch, and for the validation epoch if validation is performed. Validation result keys are prefixed with val_. For training epoch, the values of the Model's metrics are returned.

This means that if you compiled the model e.g. like this:

model.compile(loss=..., optimizer=...,
              metrics=['accuracy'])

you'd have the variables epoch, loss, val_loss, accuracy and val_accuracy available (given that validation is performed).

To have an overview of the formatting options for these values, you can then refer to the Format Specification Mini-Language of Python.

Understanding filepath naming

1 Answers1