brainspy.algorithms.gd

File containing the gradient descent algorithm methods adapted for DNPU classes and custom torch.nn.Module children custom classes that contain DNPU classes or DNPU based modules from brainspy.processors.modules.

Functions

default_train_step(model, epoch, dataloader, ...)	Deafult training step for training a torch model in Gradiet descent.
default_val_step(epoch, model, dataloader, ...)	To calulate the validation loss in each training step of the Gradient descent.
train(model, dataloaders, criterion, ...[, ...])	Main training loop for off-chip gradient descent training with early stopping using PyTorch.
train_checks(model, dataloaders, criterion, ...)	Main training loop for off-chip gradient descent training with early stopping using PyTorch.