pxd.background_generator.models Namespace Reference

basf2 (Belle II Analysis Software Framework) # Author: The Belle II Collaboration # # See git log for contributors and copyright holders. More...

Functions
Type	_get_model_cls (str model)
	Helper function that imports and returns the class Model from the Python module named as model and assumed to be located in the package pxd.background_generator.models.
Callable	_get_generate_func (str model)
	Helper function that imports and returns the function generate from the Python module named as model and assumed to be located in the package pxd.background_generator.models.

Variables
tuple	MODELS
	Container for names of generator models that are available for selection.

Detailed Description

basf2 (Belle II Analysis Software Framework) # Author: The Belle II Collaboration # # See git log for contributors and copyright holders.

# This file is licensed under LGPL-3.0, see LICENSE.md. #

This package encapsulates different generator models that can be selected to generate digits in the PXD background generator module.

Function Documentation

_get_generate_func()

Callable _get_generate_func ( str  model )

protected

Helper function that imports and returns the function generate from the Python module named as model and assumed to be located in the package pxd.background_generator.models.

An error is raised if the Python module named as model:

• does not exist,
• does not contain the function generate.

Parameters

model

Model name and the name of the corresponding Python module implementing the model

Definition at line 55 of file __init__.py.

def _get_generate_func(model: str) -> Callable:
    func = None
    try:
        func = import_module(f".{model}", __name__).generate
    except AttributeError as exc:
        exc.msg = f"please define the generation function for {model!r}"
        raise
    except ModuleNotFoundError as exc:
        exc.msg = f"the selected model {model!r} is not implemented"
        raise
    return func
 
 

_get_model_cls()

Type _get_model_cls ( str  model )

protected

Helper function that imports and returns the class Model from the Python module named as model and assumed to be located in the package pxd.background_generator.models.

An error is raised if the Python module named as model:

• does not exist,
• does not contain the class Model.

Parameters

model

Model name and the name of the corresponding Python module implementing the model

Definition at line 31 of file __init__.py.

def _get_model_cls(model: str) -> Type:
    cls = None
    try:
        cls = import_module(f".{model}", __name__).Model
    except AttributeError as exc:
        exc.msg = f"please define a model class for {model!r}"
        raise
    except ModuleNotFoundError as exc:
        exc.msg = f"the selected model {model!r} is not implemented"
        raise
    return cls
 
 

Variable Documentation

MODELS

tuple MODELS

Initial value:

=  (
    "convnet",
    "resnet",
    "ieagan"
)

Container for names of generator models that are available for selection.

Each model name is assumed to match a corresponding Python module that is located in the package pxd.background_generator.models and implements the specific model, i.e. defines a:

• class Model,
• function generate.

The class Model must inherit from the base class torch.nn.Module.

The generation function generate specific to a model is assumed to act on a model instance and produce an output of type torch.Tensor that is of shape (40, 250, 768) and contains values of type torch.uint8 in the interval . Each index in the first axis of the output tensor is assumed to correspond to the PXD sensor specified by the VxdID arguments in pxd.background_generator.VXDID_ARGS at the same index.

Definition at line 86 of file __init__.py.