graphorge.gnn_base_model.train.training

Training of Graph Neural Network model.

Classes

EarlyStopper

Early stopping procedure (implicit regularizaton).

Functions

train_model

Training of Graph Neural Network model.

get_pytorch_optimizer

Get PyTorch optimizer.

get_learning_rate_scheduler

Get PyTorch optimizer learning rate scheduler.

save_training_state

Save model and optimizer states at given training epoch.

load_training_state

Load model and optimizer states from available training data.

remove_posterior_optim_state_files

Delete optimizer training epoch state files posterior to given epoch.

save_loss_history

Save training process loss history record.

load_loss_history

Load training process loss history record.

load_lr_history

Load training process learning rate history record.

seed_worker

Set workers seed in PyTorch data loaders to preserve reproducibility.

read_loss_history_from_file

Read training loss history from loss history record file.

read_lr_history_from_file(loss_record_path)

Read training learning rate history from loss history record file.

write_training_summary_file

Write summary data file for model training process.

Functions

get_learning_rate_scheduler(optimizer, ...)	Get PyTorch optimizer learning rate scheduler.
get_pytorch_loss(loss_type, **kwargs)	Get PyTorch loss function.
get_pytorch_optimizer(algorithm, params, ...)	Get PyTorch optimizer.
load_loss_history(model, loss_nature, loss_type)	Load training process training loss history record.
load_lr_history(model[, epoch])	Load training process learning rate history record.
load_training_state(model, opt_algorithm, ...)	Load model and optimizer states from available training data.
predict(dataset, model_directory[, model, ...])	Make predictions with Graph Neural Network model for given dataset.
read_loss_history_from_file(loss_record_path)	Read training process loss history from loss history record file.
read_lr_history_from_file(loss_record_path)	Read training learning rate history from loss history record file.
remove_posterior_optim_state_files(model, epoch)	Delete optimizer training epoch state files posterior to given epoch.
save_loss_history(model, n_max_epochs, ...)	Save training process loss history record.
save_training_state(model, optimizer[, ...])	Save model and optimizer states at given training epoch.
seed_worker(worker_id)	Set workers seed in PyTorch data loaders to preserve reproducibility.
train_model(n_max_epochs, dataset, ...[, ...])	Training of Graph Neural Network model.
write_summary_file(summary_directory[, ...])	Write summary data file with provided keyword-based parameters.
write_training_summary_file(device_type, ...)	Write summary data file for model training process.

Classes

EarlyStopper(validation_dataset[, ...])	Early stopping procedure (implicit regularizaton).
GNNEPDBaseModel(n_node_in, n_node_out, ...)	GNN Encoder-Processor-Decoder base model.