Callback#

class braintools.trainer.Callback#

Base class for callbacks.

Callbacks allow you to hook into various stages of the training process. Override the methods you need in your custom callback subclass.

Examples

>>> import numpy as np
>>> import jax.numpy as jnp
>>> import brainstate
>>> import braintools
>>> class Regressor(braintools.trainer.LightningModule):
...     def __init__(self):
...         super().__init__()
...         self.linear = brainstate.nn.Linear(2, 1)
...     def training_step(self, batch, batch_idx):
...         x, y = batch
...         loss = jnp.mean((self.linear(x) - y) ** 2)
...         self.log('loss', loss, prog_bar=True)
...         return {'loss': loss}
...     def validation_step(self, batch, batch_idx):
...         x, y = batch
...         return {'loss': jnp.mean((self.linear(x) - y) ** 2)}
...     def configure_optimizers(self):
...         return braintools.optim.Adam(lr=0.1)
>>> rng = np.random.default_rng(0)
>>> X = rng.normal(size=(64, 2)).astype(np.float32)
>>> y = X @ np.array([[1.0], [-2.0]], dtype=np.float32) + 0.5
>>> loader = braintools.trainer.DataLoader((X, y), batch_size=32)
>>> brainstate.random.seed(0)
>>> model = Regressor()
>>> class EpochPrinter(braintools.trainer.Callback):
...     def on_train_epoch_end(self, trainer, module):
...         print(f"Epoch {module.current_epoch} finished")
>>> trainer = braintools.trainer.Trainer(
...     max_epochs=2, callbacks=[EpochPrinter()],
...     logger=False, enable_progress_bar=False, enable_checkpointing=False)
>>> trainer.fit(model, loader)
Epoch 0 finished
Epoch 1 finished
load_state_dict(state_dict)[source]#

Load state from state dict.

on_after_backward(trainer, module)[source]#

Called after backward pass (gradient computation).

on_after_optimizer_step(trainer, module, optimizer)[source]#

Called after each optimizer step.

on_before_backward(trainer, module, loss)[source]#

Called before backward pass (gradient computation).

on_before_optimizer_step(trainer, module, optimizer)[source]#

Called before each optimizer step.

on_fit_end(trainer, module)[source]#

Called at the very end of fit.

on_fit_start(trainer, module)[source]#

Called at the very beginning of fit.

on_load_checkpoint(trainer, module, checkpoint)[source]#

Called when loading a checkpoint.

on_predict_batch_end(trainer, module, outputs, batch, batch_idx)[source]#

Called at the end of each predict batch.

on_predict_batch_start(trainer, module, batch, batch_idx)[source]#

Called at the beginning of each predict batch.

on_predict_end(trainer, module)[source]#

Called at the end of prediction.

on_predict_start(trainer, module)[source]#

Called at the beginning of prediction.

on_save_checkpoint(trainer, module, checkpoint)[source]#

Called when saving a checkpoint.

on_test_batch_end(trainer, module, outputs, batch, batch_idx)[source]#

Called at the end of each test batch.

on_test_batch_start(trainer, module, batch, batch_idx)[source]#

Called at the beginning of each test batch.

on_test_end(trainer, module)[source]#

Called at the end of testing.

on_test_epoch_end(trainer, module)[source]#

Called at the end of each test epoch.

on_test_epoch_start(trainer, module)[source]#

Called at the beginning of each test epoch.

on_test_start(trainer, module)[source]#

Called at the beginning of testing.

on_train_batch_end(trainer, module, outputs, batch, batch_idx)[source]#

Called at the end of each training batch.

on_train_batch_start(trainer, module, batch, batch_idx)[source]#

Called at the beginning of each training batch.

on_train_end(trainer, module)[source]#

Called at the end of training.

on_train_epoch_end(trainer, module)[source]#

Called at the end of each training epoch.

on_train_epoch_start(trainer, module)[source]#

Called at the beginning of each training epoch.

on_train_start(trainer, module)[source]#

Called at the beginning of training.

on_validation_batch_end(trainer, module, outputs, batch, batch_idx)[source]#

Called at the end of each validation batch.

on_validation_batch_start(trainer, module, batch, batch_idx)[source]#

Called at the beginning of each validation batch.

on_validation_end(trainer, module)[source]#

Called at the end of validation.

on_validation_epoch_end(trainer, module)[source]#

Called at the end of each validation epoch.

on_validation_epoch_start(trainer, module)[source]#

Called at the beginning of each validation epoch.

on_validation_start(trainer, module)[source]#

Called at the beginning of validation.

state_dict()[source]#

Return state dict for checkpointing.

Return type:

Dict[str, Any]

property state_key: str#

Identifier for this callback in state dict.