KernelShap
pnpxai.explainers.kernel_shap
KernelShap
Bases: Explainer
KernelSHAP explainer.
Supported Modules: Linear, Convolution, LSTM, RNN, Attention
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
model
|
Module
|
The PyTorch model for which attribution is to be computed. |
required |
n_samples
|
int
|
Number of samples |
25
|
baseline_fn
|
Union[BaselineMethodOrFunction, Tuple[BaselineMethodOrFunction]]
|
The baseline function, accepting the attribution input, and returning the baseline accordingly. |
'zeros'
|
feature_mask_fn
|
Union[FeatureMaskMethodOrFunction, Tuple[FeatureMaskMethodOrFunction]
|
The feature mask function, accepting the attribution input, and returning the feature mask accordingly. |
'felzenszwalb'
|
mask_token_id
|
Optional[int]
|
The token id of the mask, used for modalities, utilizing tokenization |
None
|
forward_arg_extractor
|
Optional[ForwardArgumentExtractor]
|
A function that extracts forward arguments from the input batch(s) where the attribution scores are assigned. |
None
|
additional_forward_arg_extractor
|
Optional[ForwardArgumentExtractor]
|
A secondary function that extract additional forward arguments from the input batch(s). |
None
|
**kwargs
|
Keyword arguments that are forwarded to the base implementation of the Explainer |
required |
Reference
Scott M. Lundberg, Su-In Lee. A Unified Approach to Interpreting Model Predictions.
SUPPORTED_MODULES = [Linear, Convolution, LSTM, RNN, Attention]
class-attribute
instance-attribute
n_samples = n_samples
instance-attribute
baseline_fn = baseline_fn
instance-attribute
feature_mask_fn = feature_mask_fn
instance-attribute
mask_token_id = mask_token_id
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: Module, n_samples: int = 25, baseline_fn: Union[BaselineMethodOrFunction, Tuple[BaselineMethodOrFunction]] = 'zeros', feature_mask_fn: Union[FeatureMaskMethodOrFunction, Tuple[FeatureMaskMethodOrFunction]] = 'felzenszwalb', forward_arg_extractor: Optional[ForwardArgumentExtractor] = None, additional_forward_arg_extractor: Optional[ForwardArgumentExtractor] = None, mask_token_id: Optional[int] = None) -> None
attribute(inputs: Tensor, targets: Optional[Tensor] = None) -> Union[Tensor, Tuple[Tensor]]
Computes attributions for the given inputs and targets.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
inputs
|
Tensor
|
The input data. |
required |
targets
|
Tensor
|
The target labels for the inputs. |
None
|
Returns:
| Type | Description |
|---|---|
Union[Tensor, Tuple[Tensor]]
|
Union[torch.Tensor, Tuple[torch.Tensor]]: The result of the explanation. |
get_tunables() -> Dict[str, Tuple[type, Dict]]
Provides Tunable parameters for the optimizer
Tunable parameters
n_samples (int): Value can be selected in the range of range(10, 50, 10)
baseline_fn (callable): BaselineFunction selects suitable values in accordance with the modality
feature_mask_fn (callable): FeatureMaskFunction selects suitable values in accordance with the modality