pulsatrix::ResumableTrial Class Reference

A single hyperparameter configuration's live, resumable training state – own whatever network/optimizer/dataset a concrete trial needs, and train it incrementally across multiple calls rather than all at once. More...

#include <successive_halving.hpp>

Inheritance diagram for pulsatrix::ResumableTrial:

Public Member Functions
virtual	~ResumableTrial ()=default
virtual double	TrainForEpochs (int num_epochs)=0
	Trains this trial for num_epochs additional epochs (continuing from wherever this trial's own training left off, not restarting), then returns the current validation metric (maximization convention, matching every other HPO algorithm in this campaign – a caller minimizing a loss negates it).

Detailed Description

A single hyperparameter configuration's live, resumable training state – own whatever network/optimizer/dataset a concrete trial needs, and train it incrementally across multiple calls rather than all at once.

Constructor & Destructor Documentation

~ResumableTrial()

virtual pulsatrix::ResumableTrial::~ResumableTrial ( )

virtualdefault

Member Function Documentation

TrainForEpochs()

virtual double pulsatrix::ResumableTrial::TrainForEpochs ( int  num_epochs )

pure virtual

Trains this trial for num_epochs additional epochs (continuing from wherever this trial's own training left off, not restarting), then returns the current validation metric (maximization convention, matching every other HPO algorithm in this campaign – a caller minimizing a loss negates it).

---

The documentation for this class was generated from the following file:

• include/pulsatrix/successive_halving.hpp