Wraps an mlr3 Learner or GraphLearner in a vetiver model object for versioning and deployment. Vetiver provides model versioning, API generation, and monitoring capabilities.
Usage
export_vetiver(
learner,
model_name,
task,
description = NULL,
metadata = NULL,
board = NULL
)Arguments
- learner
A trained mlr3 Learner or GraphLearner
- model_name
Name for the vetiver model (required)
- task
The mlr3 Task used for training (for metadata extraction)
- description
Optional description string
- metadata
Optional list of additional metadata
- board
Optional pins board for immediate deployment. If NULL, returns the vetiver model without pinning.
Examples
if (FALSE) { # \dontrun{
library(vetiver)
library(pins)
# Train model
task <- tsk("iris")
learner <- lrn("classif.ranger", predict_type = "prob")
learner$train(task)
# Export as vetiver
v <- export_vetiver(
learner = learner,
model_name = "iris_classifier",
task = task,
description = "Random Forest classifier for iris dataset"
)
# Pin to a board
board <- board_folder("models", versioned = TRUE)
vetiver_pin_write(board, v)
# Or export and pin in one step
export_vetiver(learner, "iris_classifier", task, board = board)
} # }