Fine-tune from mlr3torch Checkpoint

Loads a checkpoint and trains the learner on a new task, attempting to warm-start when supported by the mlr3torch learner.

Usage

finetune_mlr3torch_checkpoint(learner, task, path, epochs = NULL,
  seed = NULL, device = NULL, strict = TRUE)

Arguments

learner

mlr3 learner or GraphLearner.

task

mlr3 Task to train on.

path

Checkpoint path.

epochs

Optional number of epochs to set before training.

seed

Optional seed.

device

Optional device ("cpu" or "cuda") if supported by learner.

strict

Logical, enforce exact layer matching.

Value

Trained learner.