Package a trained model for sharing — sg_package_model • segmantR

Creates a .segmantR archive (ZIP format) containing model weights, a model_card.json, and optionally training data.

Usage

sg_package_model(
  trained_model,
  output_path,
  name = NULL,
  description = NULL,
  include_training_data = FALSE,
  format = c("segmantR", "cellpose", "both")
)

Arguments

trained_model: An sg_trained_model object.
output_path: Character. Path for the output archive file.
name: Character or NULL. Human-readable model name.
description: Character or NULL. Short description of the model.
include_training_data: Logical. Include training data in the archive? Default FALSE.
format: Character. Archive format: "segmantR" (default), "cellpose", or "both".

Value

The output file path, returned invisibly.

Examples

# \donttest{
mdl <- new_sg_trained_model(
  model_path = tempdir(),
  backend = "cellpose",
  base_model = "cyto3",
  training_metrics = list(n_epochs = 50L)
)
out <- sg_package_model(mdl, tempfile(fileext = ".segmantR"))
#> ✔ Model packaged to /tmp/RtmpNfLzCr/file217c41d8416c.segmantR.
# }