Module explainability 

Explainability kernels for model interpretation.

This module provides GPU-accelerated explainability algorithms:

• SHAPValues - Kernel SHAP approximation for feature importance
• FeatureImportance - Permutation-based feature importance

Structs

FeatureImportance

Permutation Feature Importance kernel.

FeatureImportanceConfig

Configuration for permutation feature importance.

FeatureImportanceResult

Feature importance result.

SHAPBatchResult

Batch SHAP results.

SHAPConfig

Configuration for SHAP computation.

SHAPExplanation

SHAP explanation for a single prediction.

SHAPValues

SHAP Values kernel.

Enums

ImportanceMetric

Metric for measuring importance.