alibi.api.defaults
This module defines the default metadata and data dictionaries for each explanation method. Note that the "name" field is automatically populated upon initialization of the corresponding Explainer class.
Constants
DEFAULT_META_ANCHOR
DEFAULT_META_ANCHORDEFAULT_META_ANCHOR: dict = {'name': None, 'type': ['blackbox'], 'explanations': ['local'], 'params': {},...DEFAULT_DATA_ANCHOR
DEFAULT_DATA_ANCHORDEFAULT_DATA_ANCHOR: dict = {'anchor': [], 'precision': None, 'coverage': None, 'raw': None}DEFAULT_DATA_ANCHOR_IMG
DEFAULT_DATA_ANCHOR_IMGDEFAULT_DATA_ANCHOR_IMG: dict = {'anchor': [], 'segments': None, 'precision': None, 'coverage': None, 'raw': ...DEFAULT_META_CEM
DEFAULT_META_CEMDEFAULT_META_CEM: dict = {'name': None, 'type': ['blackbox', 'tensorflow', 'keras'], 'explanations': [...DEFAULT_DATA_CEM
DEFAULT_DATA_CEMDEFAULT_DATA_CEM: dict = {'PN': None, 'PP': None, 'PN_pred': None, 'PP_pred': None, 'grads_graph': Non...DEFAULT_META_CF
DEFAULT_META_CFDEFAULT_META_CF: dict = {'name': None, 'type': ['blackbox', 'tensorflow', 'keras'], 'explanations': [...DEFAULT_DATA_CF
DEFAULT_DATA_CFDEFAULT_DATA_CF: dict = {'cf': None, 'all': [], 'orig_class': None, 'orig_proba': None, 'success': None}DEFAULT_META_CFP
DEFAULT_META_CFPDEFAULT_META_CFP: dict = {'name': None, 'type': ['blackbox', 'tensorflow', 'keras'], 'explanations': [...DEFAULT_DATA_CFP
DEFAULT_DATA_CFPDEFAULT_DATA_CFP: dict = {'cf': None, 'all': [], 'orig_class': None, 'orig_proba': None, 'id_proto': N...KERNEL_SHAP_PARAMS
KERNEL_SHAP_PARAMSKERNEL_SHAP_PARAMS: list = ['link', 'group_names', 'grouped', 'groups', 'weights', 'summarise_background...Built-in mutable sequence.
If no argument is given, the constructor creates a new empty list. The argument must be an iterable if specified.
DEFAULT_META_KERNEL_SHAP
DEFAULT_META_KERNEL_SHAPDEFAULT_META_KERNEL_SHAP: dict = {'name': None, 'type': ['blackbox'], 'task': None, 'explanations': ['local', ...DEFAULT_DATA_KERNEL_SHAP
DEFAULT_DATA_KERNEL_SHAPDEFAULT_DATA_KERNEL_SHAP: dict = {'shap_values': [], 'expected_value': [], 'categorical_names': {}, 'feature_n...DEFAULT_META_ALE
DEFAULT_META_ALEDEFAULT_META_ALE: dict = {'name': None, 'type': ['blackbox'], 'explanations': ['global'], 'params': {}...DEFAULT_DATA_ALE
DEFAULT_DATA_ALEDEFAULT_DATA_ALE: dict = {'ale_values': [], 'constant_value': None, 'ale0': [], 'feature_values': [], ...TREE_SHAP_PARAMS
TREE_SHAP_PARAMSTREE_SHAP_PARAMS: list = ['model_output', 'summarise_background', 'summarise_result', 'approximate', '...Built-in mutable sequence.
If no argument is given, the constructor creates a new empty list. The argument must be an iterable if specified.
DEFAULT_META_TREE_SHAP
DEFAULT_META_TREE_SHAPDEFAULT_META_TREE_SHAP: dict = {'name': None, 'type': ['whitebox'], 'task': None, 'explanations': ['local', ...DEFAULT_DATA_TREE_SHAP
DEFAULT_DATA_TREE_SHAPDEFAULT_DATA_TREE_SHAP: dict = {'shap_values': [], 'shap_interaction_values': [], 'expected_value': [], 'cat...DEFAULT_META_INTGRAD
DEFAULT_META_INTGRADDEFAULT_META_INTGRAD: dict = {'name': None, 'type': ['whitebox'], 'explanations': ['local'], 'params': {},...DEFAULT_DATA_INTGRAD
DEFAULT_DATA_INTGRADDEFAULT_DATA_INTGRAD: dict = {'attributions': None, 'X': None, 'forward_kwargs': None, 'baselines': None, ...DEFAULT_META_CFRL
DEFAULT_META_CFRLDEFAULT_META_CFRL: dict = {'name': None, 'type': ['blackbox'], 'explanations': ['local'], 'params': {},...DEFAULT_DATA_CFRL
DEFAULT_DATA_CFRLDEFAULT_DATA_CFRL: dict = {'orig': None, 'cf': None, 'target': None, 'condition': None}DEFAULT_META_SIM
DEFAULT_META_SIMDEFAULT_META_SIM: dict = {'name': None, 'type': ['whitebox'], 'explanations': ['local'], 'params': {},...DEFAULT_DATA_SIM
DEFAULT_DATA_SIMDEFAULT_DATA_SIM: dict = {'scores': None, 'ordered_indices': None, 'most_similar': None, 'least_simila...DEFAULT_META_PROTOSELECT
DEFAULT_META_PROTOSELECTDEFAULT_META_PROTOSELECT: dict = {'name': None, 'type': ['data'], 'explanation': ['global'], 'params': {}, 've...DEFAULT_DATA_PROTOSELECT
DEFAULT_DATA_PROTOSELECTDEFAULT_DATA_PROTOSELECT: dict = {'prototypes': None, 'prototype_indices': None, 'prototype_labels': None}DEFAULT_META_PD
DEFAULT_META_PDDEFAULT_META_PD: dict = {'name': None, 'type': ['blackbox'], 'explanations': ['global'], 'params': {}...DEFAULT_DATA_PD
DEFAULT_DATA_PDDEFAULT_DATA_PD: dict = {'feature_deciles': None, 'pd_values': None, 'ice_values': None, 'feature_val...DEFAULT_META_PDVARIANCE
DEFAULT_META_PDVARIANCEDEFAULT_META_PDVARIANCE: dict = {'name': None, 'type': ['blackbox'], 'explanations': ['global'], 'params': {}...DEFAULT_DATA_PDVARIANCE
DEFAULT_DATA_PDVARIANCEDEFAULT_DATA_PDVARIANCE: dict = {'feature_deciles': None, 'pd_values': None, 'feature_values': None, 'feature...DEFAULT_META_PERMUTATION_IMPORTANCE
DEFAULT_META_PERMUTATION_IMPORTANCEDEFAULT_META_PERMUTATION_IMPORTANCE: dict = {'name': None, 'type': ['blackbox'], 'explanations': ['global'], 'params': {}...DEFAULT_DATA_PERMUTATION_IMPORTANCE
DEFAULT_DATA_PERMUTATION_IMPORTANCEDEFAULT_DATA_PERMUTATION_IMPORTANCE: dict = {'feature_names': None, 'metric_names': None, 'feature_importance': None}Last updated
Was this helpful?