Below is a list of the predefined callbacks that are available in
mlr3torch. Weight is the order in which a
callback is called within a stage: a callback with a higher weight is
called after one with a lower weight, see the section Ordering
of CallbackSet.
| Key | Label | Weight | Packages |
|---|---|---|---|
| checkpoint | Checkpoint | Inf | torch |
| history | History | 0 | torch |
| lr_cosine_annealing | Cosine Annealing LR Scheduler | 0 | torch |
| lr_lambda | Multiplication by Function LR Scheduler | 0 | torch |
| lr_multiplicative | Multiplication by Factor LR Scheduler | 0 | torch |
| lr_one_cycle | 1cycle LR Scheduler | 0 | torch |
| lr_reduce_on_plateau | Reduce on Plateau LR Scheduler | 0 | torch |
| lr_step | Step Decay LR Scheduler | 0 | torch |
| progress | Progress | 0 | progress, torch |
| tb | TensorBoard | 0 | tfevents, torch |
| unfreeze | Unfreeze | 0 | torch |