Metrics
Metrics can be passed to the Workout initializer and provide extra insights next to the loss value. Every metric is invoked with two parameters: predicted outcome and the labels.
Missing docstring for BinaryAccuracy. Check Documenter's build log for details.
Missing docstring for OneHotBinaryAccuracy. Check Documenter's build log for details.
Missing docstring for SmartReducer. Check Documenter's build log for details.
Callbacks
Callbacks can be passed to the train! function and provide a way to extend the functionality of Photon. Their functionality ranges from callbacks that generate output (like console or plots), to callbacks that save the model and callbacks that abort the training due to lack of progress.
Missing docstring for Meter. Check Documenter's build log for details.
Missing docstring for ConsoleMeter. Check Documenter's build log for details.
Missing docstring for SilentMeter. Check Documenter's build log for details.
Missing docstring for FileMeter. Check Documenter's build log for details.
Missing docstring for PlotMeter. Check Documenter's build log for details.
Missing docstring for TensorBoardMeter. Check Documenter's build log for details.
Missing docstring for EarlyStop. Check Documenter's build log for details.
Missing docstring for AutoSave. Check Documenter's build log for details.
Missing docstring for EpochSave. Check Documenter's build log for details.
Utils
Photon comes with a number of utility functions for common tasks with regards to using metrics.
Missing docstring for getmetricname. Check Documenter's build log for details.
Missing docstring for history. Check Documenter's build log for details.
Missing docstring for plotmetrics. Check Documenter's build log for details.