medkit.training.callbacks#
Classes#
A TrainerCallback is the base class for trainer callbacks. |
|
Default implementation of |
Module Contents#
- class medkit.training.callbacks.TrainerCallback#
A TrainerCallback is the base class for trainer callbacks.
- on_train_begin(config: medkit.training.trainer_config.TrainerConfig)#
Event called at the beginning of training.
- on_train_end()#
Event called at the end of training.
- on_epoch_begin(epoch: int)#
Event called at the beginning of an epoch.
- on_epoch_end(metrics: dict[str, float], epoch: int, epoch_time: float)#
Event called at the end of an epoch.
- on_step_begin(step_idx: int, nb_batches: int, phase: str)#
Event called at the beginning of a step in training.
- on_step_end(step_idx: int, nb_batches: int, phase: str)#
Event called at the end of a step in training.
- on_save(checkpoint_dir: str)#
Event called on saving a checkpoint.
- class medkit.training.callbacks.DefaultPrinterCallback#
Bases:
TrainerCallback
Default implementation of
TrainerCallback
.- logger#
- console_handler#
- formatter#
- _progress_bar = None#
- on_train_begin(config)#
Event called at the beginning of training.
- on_epoch_end(metrics, epoch, epoch_duration)#
Event called at the end of an epoch.
- on_train_end()#
Event called at the end of training.
- on_save(checkpoint_dir)#
Event called on saving a checkpoint.
- on_step_begin(step_idx: int, nb_batches: int, phase: str)#
Event called at the beginning of a step in training.
- on_step_end(step_idx: int, nb_batches: int, phase: str)#
Event called at the end of a step in training.