pihnn.utils.train#
- pihnn.utils.train(boundary, model, n_epochs, learn_rate=0.001, scheduler_apply=[], scheduler_gamma=0.5, dir='results/')#
Performs the training of the neural network.
- Parameters:
boundary (
pihnn.geometries.boundary) – Domain boundary, needed to extract training and test points.model (
pihnn.nn.PIHNN/pihnn.nn.DD_PIHNN) – Neural network model.n_epochs (int) – Number of total epochs.
learn_late – Initial learning rate for the optimizer.
scheduler_apply (list of int) – At which epoch to apply the
torch.optim.lr_scheduler.ExponentialLRscheduler.scheduler_gamma (float) – Scheduler exponential rate.
dir (str) – Directory where to save outputs.
- Returns:
loss_epochs (list of float) - List containing the training loss at each epoch.
loss_epochs_test (list of float) - List containing the test loss at each epoch.