Learn¶

Layers¶

synthtorch.learn.layers

define auxillary layers for defining neural networks in pytorch

Author: Jacob Reinhold (jacob.reinhold@jhu.edu)

Created on: Feb 21, 2018

synthtorch.learn.learner

train functions for synthtorch neural networks

Author: Jacob Reinhold (jacob.reinhold@jhu.edu)

Created on: Feb 25, 2018

synthtorch.learn.learner.get_data_augmentation(config: synthtorch.util.config.ExperimentConfig)¶: get all data augmentation transforms for training

synthtorch.learn.learner.get_dataloader(config: synthtorch.util.config.ExperimentConfig, tfms: Tuple[List, List] = None)¶: get the dataloaders for training/validation

synthtorch.learn.learner.get_device(disable_cuda=False)¶: get the device(s) for tensors to be put on

synthtorch.learn.learner.get_model(config: synthtorch.util.config.ExperimentConfig, enable_dropout: bool = True, inplace: bool = False)¶

instantiate a model based on an ExperimentConfig class instance

Parameters:	config (ExperimentConfig) – instance of the ExperimentConfig class enable_dropout (bool) – enable dropout in the model (usually for training)
Returns:	instance of one of the available models in the synthtorch package
Return type:	model

synthtorch.learn.loss

define loss functions for neural network training

Author: Jacob Reinhold (jacob.reinhold@jhu.edu)

Created on: Feb 20, 2018

synthtorch.learn.optim

define optimizer auxillary functions for neural network training

Author: Jacob Reinhold (jacob.reinhold@jhu.edu)

Created on: Feb 4, 2018

synthtorch.learn.predict

routines specific to prediction

Author: Jacob Reinhold (jacob.reinhold@jhu.edu)

Created on: Feb 26, 2018