Create an Ensemble Module — ensemble

Factory function to create an Ensemble module that combines multiple modules using a reduce function.

Usage

ensemble(modules, reduce_fn = NULL, weights = NULL, ...)

Arguments

modules: A list of Module objects to combine
reduce_fn: Function to combine outputs. Default is reduce_majority(). Should accept a list of outputs and optional weights parameter.
weights: Optional numeric vector of weights for each module. Useful for weighted voting based on validation performance.
...: Additional arguments passed to module constructor

Value

An EnsembleModule object

Examples

# Create multiple compiled modules
sig <- signature("question -> answer")
mod1 <- module(sig, type = "predict")
mod2 <- module(sig, type = "predict")
mod3 <- module(sig, type = "predict")

# Combine with majority voting
ens <- ensemble(list(mod1, mod2, mod3))

# With weighted voting based on validation scores
ens <- ensemble(
  list(mod1, mod2, mod3),
  reduce_fn = reduce_weighted_vote(),
  weights = c(0.9, 0.85, 0.8)
)

# With custom reduce function
custom_reduce <- function(outputs, weights = NULL) {
  # Take the longest answer
  lengths <- vapply(outputs, function(o) nchar(o$answer %||% ""), integer(1))
  outputs[[which.max(lengths)]]
}
ens <- ensemble(list(mod1, mod2, mod3), reduce_fn = custom_reduce)