hookeai.model_architectures.procedures.model_training.save_training_state

save_training_state(model, optimizer, state_type, epoch=None, is_remove_posterior=True)

Save model and optimizer states at given training epoch.

Material patch model state file is stored in model_directory under the name < model_name >.pt or < model_name >-< epoch >.pt if epoch is known.

Material patch model state file corresponding to the best performance is stored in model_directory under the name < model_name >-best.pt or < model_name >-< epoch >-best.pt if epoch is known.

Optimizer state file is stored in model_directory under the name < model_name >_optim-< epoch >.pt.

Optimizer state file corresponding to the best performance is stored in model_directory under the name < model_name >_optim-best.pt or < model_name >_optim-< epoch >-best.pt if epoch is known.

Parameters:

• model (torch.nn.Module) – Model.

• optimizer (torch.optim.Optimizer) – PyTorch optimizer.

• state_type ({'default', 'init', 'epoch', 'best'}, default='default') –

  Saved model state file type. Options:

  ’default’ : Model default state

  ’init’ : Model initial state

  ’epoch’ : Model state of given training epoch

  ’best’ : Model state of best performance

• epoch (int, default=None) – Training epoch.

• is_remove_posterior (bool, default=True) – Remove material patch model and optimizer state files corresponding to training epochs posterior to the saved state file. Effective only if saved epoch is known.