When passing the rlang::expr of estimators to origami::cross_validate with the parallel option set to TRUE, the underlying call to future_lapply would evaluate the estimators expr. This would then cause errors in the fold function. I'm not exactly sure why this occurs, but it was fixed by capturing the collection of estimators as a rlang::quo instead.
When passing the
rlang::exprof estimators toorigami::cross_validatewith the parallel option set toTRUE, the underlying call tofuture_lapplywould evaluate the estimatorsexpr. This would then cause errors in the fold function. I'm not exactly sure why this occurs, but it was fixed by capturing the collection of estimators as arlang::quoinstead.