Last chance! 50% off unlimited learning
Sale ends in
This is conceptually similar to zip() in Python, or R functions
purrr::transpose() and data.table::transpose() (albeit, accepting
elements in ... instead of a single list), with one crucial difference: if
the provided objects are named, then matching is done by names, not
positions.
zip_lists(...)A inverted list
R lists or atomic vectors, optionally named.
All arguments supplied must be of the same length. If positional matching is required, then all arguments provided must be unnamed. If matching by names, then all arguments must have the same set of names, but they can be in different orders.
gradients <- list("grad_for_wt_1", "grad_for_wt_2", "grad_for_wt_3")
weights <- list("weight_1", "weight_2", "weight_3")
str(zip_lists(gradients, weights))
str(zip_lists(gradient = gradients, weight = weights))
names(gradients) <- names(weights) <- paste0("layer_", 1:3)
str(zip_lists(gradients, weights[c(3, 1, 2)]))
names(gradients) <- paste0("gradient_", 1:3)
try(zip_lists(gradients, weights)) # error, names don't match
# call unname directly for positional matching
str(zip_lists(unname(gradients), unname(weights)))
Run the code above in your browser using DataLab