

Callbacks

---

Callbacks can be used with PySHAC for custom code execution, such as monitoring the improvement of the engine, maintaining a history of the training session or simply writing the logs to files.

Callbacks allow you the flexibility to prepare your state prior to training or evaluation, and in doing so can allow one to perform stateful evaluation if necessary using concurrent evaluators.

Usage

Callbacks can be imported from the pyshac.config.callbacks package as shown below.

from pyshac.config.callbacks import History, CSVLogger

shac = SHAC(...)

# History is not needed here, as it is automatically added by default for all .fit / .fit_dataset calls.
callbacks = [History(), CSVLogger('path/to/file.csv')]

history = shac.fit(evaluation_function, callbacks=callbacks)
OR
history = shac.fit_dataset('path/to/dataset', callbacks=callbacks)

print(history.history)

History

All calls to shac.fit and shac.fit_dataset will now return a History object, which is a callback to monitor and log all valuable information occurring during training.

The History object has a special member, also called history, which is a dictionary containing all of the logged values.

History is added by default

The History callback is added by default to all calls to fit or fit_dataset and therefore there it is not necessary to add this callback manually.