RAP

pnpxai.explainers.rap.attribute

RAP

Bases: Explainer

Computes Relative Attribute Propagation (RAP) explanations for a given model.

Supported Modules: Linear, Convolution

Parameters:

Name	Type	Description	Default
model	Model	The model for which RAP explanations are computed.	required

Reference

Woo-Jeoung Nam, Shir Gur, Jaesik Choi, Lior Wolf, Seong-Whan Lee. Relative Attributing Propagation: Interpreting the Comparative Contributions of Individual Units in Deep Neural Networks.

SUPPORTED_MODULES = [Linear, Convolution] class-attribute instance-attribute

method = RelativeAttributePropagation(model) instance-attribute

EXPLANATION_TYPE: ExplanationType = 'attribution' class-attribute instance-attribute

TUNABLES = {} class-attribute instance-attribute

model = model.eval() instance-attribute

forward_arg_extractor = forward_arg_extractor instance-attribute

additional_forward_arg_extractor = additional_forward_arg_extractor instance-attribute

device property

__init__(model: Model)

compute_pred(output: Tensor) -> Tensor

Computes the predicted class probabilities.

Parameters:

Name	Type	Description	Default
output	Tensor	The model output.	required

Returns:

Name	Type	Description
Tensor	Tensor	The one-hot encoded predicted class probabilities.

attribute(inputs: DataSource, targets: DataSource, *args: Any, **kwargs: Any) -> DataSource

Computes RAP attributions for the given inputs.

Parameters:

Name	Type	Description	Default
inputs	DataSource	The input data.	required
targets	DataSource	The target labels.	required
*args	Any	Additional positional arguments.	()
**kwargs	Any	Additional keyword arguments.	{}

Returns:

Name	Type	Description
DataSource	DataSource	RAP attributions.

format_outputs_for_visualization(inputs: DataSource, targets: DataSource, explanations: DataSource, task: Task, kwargs: Optional[Dict[str, Any]] = None)

__repr__()

copy()

set_kwargs(**kwargs)

get_tunables() -> Dict[str, Tuple[type, dict]]

Returns a dictionary of tunable parameters for the explainer.

Returns:

Type	Description
Dict[str, Tuple[type, dict]]	Dict[str, Tuple[type, dict]]: Dictionary of tunable parameters.