Has anyone successfully employed

shap

with the

N-HITS

exogenous variables? Or is this approach not viable? I'm hoping to attain some explainability or interpretability from the fitted

N-HITS

model incorporating these exogenous variables.