factory module#
- class factory.IGANFactory(parser: ConfigParser)[source]#
Bases:
ABC
Factory that represents a GAN. This factory does not keep of created references.
- __init__(parser: ConfigParser) None [source]#
Initialize the factory.
- Parameters:
parser (ConfigParser) – Parser for config file containing GAN model and training params.
- abstract get_trainer() Callable [source]#
Returns the GAN train function.
- Returns:
GAN train() function.
- Return type:
Callable
- _abc_impl = <_abc._abc_data object>#
- class factory.GANFactory(parser: ConfigParser)[source]#
Bases:
IGANFactory
- get_trainer() Callable [source]#
Returns the GAN train function.
- Returns:
GAN train() function.
- Return type:
Callable
- _abc_impl = <_abc._abc_data object>#
- class factory.ConditionalCatGANFactory(parser: ConfigParser)[source]#
Bases:
IGANFactory
- get_gan() ConditionalCatGAN [source]#
Returns a GAN instance
- Returns:
GAN instance.
- Return type:
- get_trainer() Callable [source]#
Returns the GAN train function.
- Returns:
GAN train() function.
- Return type:
Callable
- _abc_impl = <_abc._abc_data object>#
- class factory.ConditionalProjGANFactory(parser: ConfigParser)[source]#
Bases:
IGANFactory
- get_gan() ConditionalProjGAN [source]#
Returns a GAN instance
- Returns:
GAN instance.
- Return type:
- get_trainer() Callable [source]#
Returns the GAN train function.
- Returns:
GAN train() function.
- Return type:
Callable
- _abc_impl = <_abc._abc_data object>#
- class factory.CausalGANFactory(parser: ConfigParser)[source]#
Bases:
IGANFactory
- get_trainer() Callable [source]#
Returns the GAN train function.
- Returns:
GAN train() function.
- Return type:
Callable
- _abc_impl = <_abc._abc_data object>#
- factory.get_factory(cfg: ConfigParser) IGANFactory [source]#
Return the factory for the GAN type based on ‘model’ key in the parser.
- Parameters:
cfg (ConfigParser) – Parser for config file containing GAN model and training params.
- Returns:
Factory for the specified GAN.
- Return type:
- Raises:
ValueError – If the model is unknown or not implemented.