sakura.sakuraAE.sakuraAE.save_checkpoint

sakuraAE.save_checkpoint(training_state=None, checkpoint_path=None, save_model_arch=False, save_config=False)

Save the current state of the model and training process as a checkpoint.

Parameters:

• training_state (dict[str, Any], optional) – A dictionary containing the current state of the training process， which may include the current tick number, epoch number, data sampler status, etc.

• checkpoint_path (str, optional) – File path where the checkpoint will be saved

• save_model_arch (bool, optional) – Whether to save the model architecture in the checkpoint, defaults to False

• save_config (bool, optional) – Whether to save a redundant copy of the model’s configuration for checking, defaults to False

Returns:

None