.. testsetup:: *

    from pytorch_lightning.trainer.trainer import Trainer
    from pytorch_lightning.callbacks.base import Callback

A callback is a self-contained program that can be reused across projects.

Lightning has a callback system to execute callbacks when needed. Callbacks should capture NON-ESSENTIAL logic that is NOT required for your :ref:`lightning_module` to run.

Here's the flow of how the callback hooks are executed:

An overall Lightning system should have:

1. Trainer for all engineering
2. LightningModule for all research code.
3. Callbacks for non-essential code.

Example:

.. testcode::

    from pytorch_lightning.callbacks import Callback

    class MyPrintingCallback(Callback):

        def on_init_start(self, trainer):
            print('Starting to init trainer!')

        def on_init_end(self, trainer):
            print('trainer is init now')

        def on_train_end(self, trainer, pl_module):
            print('do something when training ends')

    trainer = Trainer(callbacks=[MyPrintingCallback()])

.. testoutput::

    Starting to init trainer!
    trainer is init now

We successfully extended functionality without polluting our super clean :ref:`lightning_module` research code.

You can do pretty much anything with callbacks.

• Add a MLP to fine-tune self-supervised networks.
• Find how to modify an image input to trick the classification result.
• Interpolate the latent space of any variational model.
• Log images to Tensorboard for any model.

Lightning has a few built-in callbacks.

.. currentmodule:: pytorch_lightning.callbacks

.. autosummary::
    :toctree: generated
    :nosignatures:
    :template: classtemplate.rst

    Callback
    EarlyStopping
    GPUStatsMonitor
    GradientAccumulationScheduler
    LearningRateMonitor
    ModelCheckpoint
    ProgressBar
    ProgressBarBase

Some callbacks require internal state in order to function properly. You can optionally choose to persist your callback's state as part of model checkpoint files using the callback hooks :meth:`~pytorch_lightning.callbacks.Callback.on_save_checkpoint` and :meth:`~pytorch_lightning.callbacks.Callback.on_load_checkpoint`. However, you must follow two constraints:

1. Your returned state must be able to be pickled.
2. You can only use one instance of that class in the Trainer callbacks list. We don't support persisting state for multiple callbacks of the same class.

The following are best practices when using/designing callbacks.

1. Callbacks should be isolated in their functionality.
2. Your callback should not rely on the behavior of other callbacks in order to work properly.
3. Do not manually call methods from the callback.
4. Directly calling methods (eg. on_validation_end) is strongly discouraged.
5. Whenever possible, your callbacks should not depend on the order in which they are executed.

.. automethod:: pytorch_lightning.callbacks.Callback.setup
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.teardown
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_init_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_init_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_save_checkpoint
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_fit_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_sanity_check_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_sanity_check_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_train_batch_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_train_batch_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_train_epoch_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_train_epoch_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_validation_epoch_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_validation_epoch_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_test_epoch_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_test_epoch_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_epoch_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_epoch_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_batch_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_validation_batch_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_validation_batch_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_test_batch_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_test_batch_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_batch_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_train_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_train_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_pretrain_routine_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_pretrain_routine_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_validation_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_validation_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_test_start
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_test_end
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_keyboard_interrupt
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_save_checkpoint
    :noindex:

.. automethod:: pytorch_lightning.callbacks.Callback.on_load_checkpoint
    :noindex: