.. role:: hidden
:class: hidden-section
.. _lightning_module:
###############
LightningModule
###############
A :class:`~LightningModule` organizes your PyTorch code into 6 sections:
- Computations (init).
- Train Loop (training_step)
- Validation Loop (validation_step)
- Test Loop (test_step)
- Prediction Loop (predict_step)
- Optimizers and LR Schedulers (configure_optimizers)
|
.. raw:: html
|
Notice a few things.
1. It is the SAME code.
2. The PyTorch code IS NOT abstracted - just organized.
3. All the other code that's not in the :class:`~LightningModule`
has been automated for you by the Trainer.
|
.. code-block:: python
net = Net()
trainer = Trainer()
trainer.fit(net)
4. There are no ``.cuda()`` or ``.to(device)`` calls required. Lightning does these for you.
|
.. code-block:: python
# don't do in Lightning
x = torch.Tensor(2, 3)
x = x.cuda()
x = x.to(device)
# do this instead
x = x # leave it alone!
# or to init a new tensor
new_x = torch.Tensor(2, 3)
new_x = new_x.to(x)
5. When running under a distributed strategy, Lightning handles the distributed sampler for you by default.
|
.. code-block:: python
# Don't do in Lightning...
data = MNIST(...)
sampler = DistributedSampler(data)
DataLoader(data, sampler=sampler)
# do this instead
data = MNIST(...)
DataLoader(data)
6. A :class:`~LightningModule` is a :class:`torch.nn.Module` but with added functionality. Use it as such!
|
.. code-block:: python
net = Net.load_from_checkpoint(PATH)
net.freeze()
out = net(x)
Thus, to use Lightning, you just need to organize your code which takes about 30 minutes,
(and let's be real, you probably should do anyway).
------------
***************
Starter Example
***************
Here are the only required methods.
.. code-block:: python
import pytorch_lightning as pl
import torch.nn as nn
import torch.nn.functional as F
class LitModel(pl.LightningModule):
def __init__(self):
super().__init__()
self.l1 = nn.Linear(28 * 28, 10)
def forward(self, x):
return torch.relu(self.l1(x.view(x.size(0), -1)))
def training_step(self, batch, batch_idx):
x, y = batch
y_hat = self(x)
loss = F.cross_entropy(y_hat, y)
return loss
def configure_optimizers(self):
return torch.optim.Adam(self.parameters(), lr=0.02)
Which you can train by doing:
.. code-block:: python
train_loader = DataLoader(MNIST(os.getcwd(), download=True, transform=transforms.ToTensor()))
trainer = pl.Trainer(max_epochs=1)
model = LitModel()
trainer.fit(model, train_dataloaders=train_loader)
The LightningModule has many convenience methods, but the core ones you need to know about are:
.. list-table::
:widths: 50 50
:header-rows: 1
* - Name
- Description
* - init
- Define computations here
* - forward
- Use for inference only (separate from training_step)
* - training_step
- the complete training loop
* - validation_step
- the complete validation loop
* - test_step
- the complete test loop
* - predict_step
- the complete prediction loop
* - configure_optimizers
- define optimizers and LR schedulers
----------
********
Training
********
Training Loop
=============
To activate the training loop, override the :meth:`~pytorch_lightning.core.module.LightningModule.training_step` method.
.. code-block:: python
class LitClassifier(pl.LightningModule):
def __init__(self, model):
super().__init__()
self.model = model
def training_step(self, batch, batch_idx):
x, y = batch
y_hat = self.model(x)
loss = F.cross_entropy(y_hat, y)
return loss
Under the hood, Lightning does the following (pseudocode):
.. code-block:: python
# put model in train mode and enable gradient calculation
model.train()
torch.set_grad_enabled(True)
outs = []
for batch_idx, batch in enumerate(train_dataloader):
loss = training_step(batch, batch_idx)
outs.append(loss.detach())
# clear gradients
optimizer.zero_grad()
# backward
loss.backward()
# update parameters
optimizer.step()
Train Epoch-level Metrics
=========================
If you want to calculate epoch-level metrics and log them, use :meth:`~pytorch_lightning.core.module.LightningModule.log`.
.. code-block:: python
def training_step(self, batch, batch_idx):
x, y = batch
y_hat = self.model(x)
loss = F.cross_entropy(y_hat, y)
# logs metrics for each training_step,
# and the average across the epoch, to the progress bar and logger
self.log("train_loss", loss, on_step=True, on_epoch=True, prog_bar=True, logger=True)
return loss
The :meth:`~pytorch_lightning.core.module.LightningModule.log` object automatically reduces the
requested metrics across a complete epoch and devices. Here's the pseudocode of what it does under the hood:
.. code-block:: python
outs = []
for batch_idx, batch in enumerate(train_dataloader):
# forward
loss = training_step(batch, batch_idx)
outs.append(loss)
# clear gradients
optimizer.zero_grad()
# backward
loss.backward()
# update parameters
optimizer.step()
epoch_metric = torch.mean(torch.stack([x for x in outs]))
Train Epoch-level Operations
============================
If you need to do something with all the outputs of each :meth:`~pytorch_lightning.core.module.LightningModule.training_step`,
override the :meth:`~pytorch_lightning.core.module.LightningModule.training_epoch_end` method.
.. code-block:: python
def training_step(self, batch, batch_idx):
x, y = batch
y_hat = self.model(x)
loss = F.cross_entropy(y_hat, y)
preds = ...
return {"loss": loss, "other_stuff": preds}
def training_epoch_end(self, training_step_outputs):
all_preds = torch.stack(training_step_outputs)
...
The matching pseudocode is:
.. code-block:: python
outs = []
for batch_idx, batch in enumerate(train_dataloader):
# forward
loss = training_step(batch, batch_idx)
outs.append(loss)
# clear gradients
optimizer.zero_grad()
# backward
loss.backward()
# update parameters
optimizer.step()
training_epoch_end(outs)
Training with DataParallel
==========================
When training using a ``strategy`` that splits data from each batch across GPUs, sometimes you might
need to aggregate them on the main GPU for processing (DP).
In this case, implement the :meth:`~pytorch_lightning.core.module.LightningModule.training_step_end`
method which will have outputs from all the devices and you can accumulate to get the effective results.
.. code-block:: python
def training_step(self, batch, batch_idx):
x, y = batch
y_hat = self.model(x)
loss = F.cross_entropy(y_hat, y)
pred = ...
return {"loss": loss, "pred": pred}
def training_step_end(self, batch_parts):
# predictions from each GPU
predictions = batch_parts["pred"]
# losses from each GPU
losses = batch_parts["loss"]
gpu_0_prediction = predictions[0]
gpu_1_prediction = predictions[1]
# do something with both outputs
return (losses[0] + losses[1]) / 2
def training_epoch_end(self, training_step_outputs):
for out in training_step_outputs:
...
Here is the Lightning training pseudo-code for DP:
.. code-block:: python
outs = []
for batch_idx, train_batch in enumerate(train_dataloader):
batches = split_batch(train_batch)
dp_outs = []
for sub_batch in batches:
# 1
dp_out = training_step(sub_batch, batch_idx)
dp_outs.append(dp_out)
# 2
out = training_step_end(dp_outs)
outs.append(out)
# do something with the outputs for all batches
# 3
training_epoch_end(outs)
------------------
**********
Validation
**********
Validation Loop
===============
To activate the validation loop while training, override the :meth:`~pytorch_lightning.core.module.LightningModule.validation_step` method.
.. code-block:: python
class LitModel(pl.LightningModule):
def validation_step(self, batch, batch_idx):
x, y = batch
y_hat = self.model(x)
loss = F.cross_entropy(y_hat, y)
self.log("val_loss", loss)
Under the hood, Lightning does the following (pseudocode):
.. code-block:: python
# ...
for batch_idx, batch in enumerate(train_dataloader):
loss = model.training_step(batch, batch_idx)
loss.backward()
# ...
if validate_at_some_point:
# disable grads + batchnorm + dropout
torch.set_grad_enabled(False)
model.eval()
# ----------------- VAL LOOP ---------------
for val_batch_idx, val_batch in enumerate(val_dataloader):
val_out = model.validation_step(val_batch, val_batch_idx)
# ----------------- VAL LOOP ---------------
# enable grads + batchnorm + dropout
torch.set_grad_enabled(True)
model.train()
You can also run just the validation loop on your validation dataloaders by overriding :meth:`~pytorch_lightning.core.module.LightningModule.validation_step`
and calling :meth:`~pytorch_lightning.trainer.trainer.Trainer.validate`.
.. code-block:: python
model = Model()
trainer = Trainer()
trainer.validate(model)
.. note::
It is recommended to validate on single device to ensure each sample/batch gets evaluated exactly once.
This is helpful to make sure benchmarking for research papers is done the right way. Otherwise, in a
multi-device setting, samples could occur duplicated when :class:`~torch.utils.data.distributed.DistributedSampler`
is used, for eg. with ``strategy="ddp"``. It replicates some samples on some devices to make sure all devices have
same batch size in case of uneven inputs.
Validation Epoch-level Metrics
==============================
If you need to do something with all the outputs of each :meth:`~pytorch_lightning.core.module.LightningModule.validation_step`,
override the :meth:`~pytorch_lightning.core.module.LightningModule.validation_epoch_end` method. Note that this method is called before :meth:`~pytorch_lightning.core.module.LightningModule.training_epoch_end`.
.. code-block:: python
def validation_step(self, batch, batch_idx):
x, y = batch
y_hat = self.model(x)
loss = F.cross_entropy(y_hat, y)
pred = ...
return pred
def validation_epoch_end(self, validation_step_outputs):
all_preds = torch.stack(validation_step_outputs)
...
Validating with DataParallel
============================
When validating using a ``strategy`` that splits data from each batch across GPUs, sometimes you might
need to aggregate them on the main GPU for processing (DP).
In this case, implement the :meth:`~pytorch_lightning.core.module.LightningModule.validation_step_end`
method which will have outputs from all the devices and you can accumulate to get the effective results.
.. code-block:: python
def validation_step(self, batch, batch_idx):
x, y = batch
y_hat = self.model(x)
loss = F.cross_entropy(y_hat, y)
pred = ...
return {"loss": loss, "pred": pred}
def validation_step_end(self, batch_parts):
# predictions from each GPU
predictions = batch_parts["pred"]
# losses from each GPU
losses = batch_parts["loss"]
gpu_0_prediction = predictions[0]
gpu_1_prediction = predictions[1]
# do something with both outputs
return (losses[0] + losses[1]) / 2
def validation_epoch_end(self, validation_step_outputs):
for out in validation_step_outputs:
...
Here is the Lightning validation pseudo-code for DP:
.. code-block:: python
outs = []
for batch in dataloader:
batches = split_batch(batch)
dp_outs = []
for sub_batch in batches:
# 1
dp_out = validation_step(sub_batch)
dp_outs.append(dp_out)
# 2
out = validation_step_end(dp_outs)
outs.append(out)
# do something with the outputs for all batches
# 3
validation_epoch_end(outs)
----------------
*******
Testing
*******
Test Loop
=========
The process for enabling a test loop is the same as the process for enabling a validation loop. Please refer to
the section above for details. For this you need to override the :meth:`~pytorch_lightning.core.module.LightningModule.test_step` method.
The only difference is that the test loop is only called when :meth:`~pytorch_lightning.trainer.trainer.Trainer.test` is used.
.. code-block:: python
model = Model()
trainer = Trainer()
trainer.fit(model)
# automatically loads the best weights for you
trainer.test(model)
There are two ways to call ``test()``:
.. code-block:: python
# call after training
trainer = Trainer()
trainer.fit(model)
# automatically auto-loads the best weights from the previous run
trainer.test(dataloaders=test_dataloader)
# or call with pretrained model
model = MyLightningModule.load_from_checkpoint(PATH)
trainer = Trainer()
trainer.test(model, dataloaders=test_dataloader)
.. note::
It is recommended to validate on single device to ensure each sample/batch gets evaluated exactly once.
This is helpful to make sure benchmarking for research papers is done the right way. Otherwise, in a
multi-device setting, samples could occur duplicated when :class:`~torch.utils.data.distributed.DistributedSampler`
is used, for eg. with ``strategy="ddp"``. It replicates some samples on some devices to make sure all devices have
same batch size in case of uneven inputs.
----------
*********
Inference
*********
Prediction Loop
===============
By default, the :meth:`~pytorch_lightning.core.module.LightningModule.predict_step` method runs the
:meth:`~pytorch_lightning.core.module.LightningModule.forward` method. In order to customize this behaviour,
simply override the :meth:`~pytorch_lightning.core.module.LightningModule.predict_step` method.
For the example let's override ``predict_step`` and try out `Monte Carlo Dropout `_:
.. code-block:: python
class LitMCdropoutModel(pl.LightningModule):
def __init__(self, model, mc_iteration):
super().__init__()
self.model = model
self.dropout = nn.Dropout()
self.mc_iteration = mc_iteration
def predict_step(self, batch, batch_idx):
# enable Monte Carlo Dropout
self.dropout.train()
# take average of `self.mc_iteration` iterations
pred = torch.vstack([self.dropout(self.model(x)).unsqueeze(0) for _ in range(self.mc_iteration)]).mean(dim=0)
return pred
Under the hood, Lightning does the following (pseudocode):
.. code-block:: python
# disable grads + batchnorm + dropout
torch.set_grad_enabled(False)
model.eval()
all_preds = []
for batch_idx, batch in enumerate(predict_dataloader):
pred = model.predict_step(batch, batch_idx)
all_preds.append(pred)
There are two ways to call ``predict()``:
.. code-block:: python
# call after training
trainer = Trainer()
trainer.fit(model)
# automatically auto-loads the best weights from the previous run
predictions = trainer.predict(dataloaders=predict_dataloader)
# or call with pretrained model
model = MyLightningModule.load_from_checkpoint(PATH)
trainer = Trainer()
predictions = trainer.predict(model, dataloaders=test_dataloader)
Inference in Research
=====================
If you want to perform inference with the system, you can add a ``forward`` method to the LightningModule.
.. note:: When using forward, you are responsible to call :func:`~torch.nn.Module.eval` and use the :func:`~torch.no_grad` context manager.
.. code-block:: python
class Autoencoder(pl.LightningModule):
def forward(self, x):
return self.decoder(x)
model = Autoencoder()
model.eval()
with torch.no_grad():
reconstruction = model(embedding)
The advantage of adding a forward is that in complex systems, you can do a much more involved inference procedure,
such as text generation:
.. code-block:: python
class Seq2Seq(pl.LightningModule):
def forward(self, x):
embeddings = self(x)
hidden_states = self.encoder(embeddings)
for h in hidden_states:
# decode
...
return decoded
In the case where you want to scale your inference, you should be using
:meth:`~pytorch_lightning.core.module.LightningModule.predict_step`.
.. code-block:: python
class Autoencoder(pl.LightningModule):
def forward(self, x):
return self.decoder(x)
def predict_step(self, batch, batch_idx, dataloader_idx=0):
# this calls forward
return self(batch)
data_module = ...
model = Autoencoder()
trainer = Trainer(accelerator="gpu", devices=2)
trainer.predict(model, data_module)
Inference in Production
=======================
For cases like production, you might want to iterate different models inside a LightningModule.
.. code-block:: python
from torchmetrics.functional import accuracy
class ClassificationTask(pl.LightningModule):
def __init__(self, model):
super().__init__()
self.model = model
def training_step(self, batch, batch_idx):
x, y = batch
y_hat = self.model(x)
loss = F.cross_entropy(y_hat, y)
return loss
def validation_step(self, batch, batch_idx):
loss, acc = self._shared_eval_step(batch, batch_idx)
metrics = {"val_acc": acc, "val_loss": loss}
self.log_dict(metrics)
return metrics
def test_step(self, batch, batch_idx):
loss, acc = self._shared_eval_step(batch, batch_idx)
metrics = {"test_acc": acc, "test_loss": loss}
self.log_dict(metrics)
return metrics
def _shared_eval_step(self, batch, batch_idx):
x, y = batch
y_hat = self.model(x)
loss = F.cross_entropy(y_hat, y)
acc = accuracy(y_hat, y)
return loss, acc
def predict_step(self, batch, batch_idx, dataloader_idx=0):
x, y = batch
y_hat = self.model(x)
return y_hat
def configure_optimizers(self):
return torch.optim.Adam(self.model.parameters(), lr=0.02)
Then pass in any arbitrary model to be fit with this task
.. code-block:: python
for model in [resnet50(), vgg16(), BidirectionalRNN()]:
task = ClassificationTask(model)
trainer = Trainer(accelerator="gpu", devices=2)
trainer.fit(task, train_dataloaders=train_dataloader, val_dataloaders=val_dataloader)
Tasks can be arbitrarily complex such as implementing GAN training, self-supervised or even RL.
.. code-block:: python
class GANTask(pl.LightningModule):
def __init__(self, generator, discriminator):
super().__init__()
self.generator = generator
self.discriminator = discriminator
...
When used like this, the model can be separated from the Task and thus used in production without needing to keep it in
a ``LightningModule``.
The following example shows how you can run inference in the Python runtime:
.. code-block:: python
task = ClassificationTask(model)
trainer = Trainer(accelerator="gpu", devices=2)
trainer.fit(task, train_dataloader, val_dataloader)
trainer.save_checkpoint("best_model.ckpt")
# use model after training or load weights and drop into the production system
model = ClassificationTask.load_from_checkpoint("best_model.ckpt")
x = ...
model.eval()
with torch.no_grad():
y_hat = model(x)
Check out :ref:`Inference in Production ` guide to learn about the possible ways to perform inference in production.
-----------
********************
Save Hyperparameters
********************
Often times we train many versions of a model. You might share that model or come back to it a few months later at which
point it is very useful to know how that model was trained (i.e.: what learning rate, neural network, etc...).
Lightning has a standardized way of saving the information for you in checkpoints and YAML files. The goal here is to
improve readability and reproducibility.
save_hyperparameters
====================
Use :meth:`~pytorch_lightning.core.module.LightningModule.save_hyperparameters` within your
:class:`~pytorch_lightning.core.module.LightningModule`'s ``__init__`` method. It will enable Lightning to store all the
provided arguments under the ``self.hparams`` attribute. These hyperparameters will also be stored within the model
checkpoint, which simplifies model re-instantiation after training.
.. code-block:: python
class LitMNIST(LightningModule):
def __init__(self, layer_1_dim=128, learning_rate=1e-2):
super().__init__()
# call this to save (layer_1_dim=128, learning_rate=1e-4) to the checkpoint
self.save_hyperparameters()
# equivalent
self.save_hyperparameters("layer_1_dim", "learning_rate")
# Now possible to access layer_1_dim from hparams
self.hparams.layer_1_dim
In addition, loggers that support it will automatically log the contents of ``self.hparams``.
Excluding hyperparameters
=========================
By default, every parameter of the ``__init__`` method will be considered a hyperparameter to the LightningModule.
However, sometimes some parameters need to be excluded from saving, for example when they are not serializable. Those
parameters should be provided back when reloading the LightningModule. In this case, exclude them explicitly:
.. code-block:: python
class LitMNIST(LightningModule):
def __init__(self, loss_fx, generator_network, layer_1_dim=128):
super().__init__()
self.layer_1_dim = layer_1_dim
self.loss_fx = loss_fx
# call this to save only (layer_1_dim=128) to the checkpoint
self.save_hyperparameters("layer_1_dim")
# equivalent
self.save_hyperparameters(ignore=["loss_fx", "generator_network"])
load_from_checkpoint
====================
LightningModules that have hyperparameters automatically saved with
:meth:`~pytorch_lightning.core.module.LightningModule.save_hyperparameters` can conveniently be loaded and instantiated
directly from a checkpoint with :meth:`~pytorch_lightning.core.module.LightningModule.load_from_checkpoint`:
.. code-block:: python
# to load specify the other args
model = LitMNIST.load_from_checkpoint(PATH, loss_fx=torch.nn.SomeOtherLoss, generator_network=MyGenerator())
If parameters were excluded, they need to be provided at the time of loading:
.. code-block:: python
# the excluded parameters were `loss_fx` and `generator_network`
model = LitMNIST.load_from_checkpoint(PATH, loss_fx=torch.nn.SomeOtherLoss, generator_network=MyGenerator())
-----------
*************
Child Modules
*************
.. include:: ../common/child_modules.rst
-----------
*******************
LightningModule API
*******************
Methods
=======
all_gather
~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.all_gather
:noindex:
configure_callbacks
~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.configure_callbacks
:noindex:
configure_optimizers
~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.configure_optimizers
:noindex:
forward
~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.forward
:noindex:
freeze
~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.freeze
:noindex:
.. _lm-log:
log
~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.log
:noindex:
log_dict
~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.log_dict
:noindex:
lr_schedulers
~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.lr_schedulers
:noindex:
manual_backward
~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.manual_backward
:noindex:
optimizers
~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.optimizers
:noindex:
print
~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.print
:noindex:
predict_step
~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.predict_step
:noindex:
save_hyperparameters
~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.save_hyperparameters
:noindex:
toggle_optimizer
~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.toggle_optimizer
:noindex:
test_step
~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.test_step
:noindex:
test_step_end
~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.test_step_end
:noindex:
test_epoch_end
~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.test_epoch_end
:noindex:
to_onnx
~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.to_onnx
:noindex:
to_torchscript
~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.to_torchscript
:noindex:
training_step
~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.training_step
:noindex:
training_step_end
~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.training_step_end
:noindex:
training_epoch_end
~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.training_epoch_end
:noindex:
unfreeze
~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.unfreeze
:noindex:
untoggle_optimizer
~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.untoggle_optimizer
:noindex:
validation_step
~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.validation_step
:noindex:
validation_step_end
~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.validation_step_end
:noindex:
validation_epoch_end
~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.validation_epoch_end
:noindex:
-----------
Properties
==========
These are properties available in a LightningModule.
current_epoch
~~~~~~~~~~~~~
The number of epochs run.
.. code-block:: python
def training_step(self, batch, batch_idx):
if self.current_epoch == 0:
...
device
~~~~~~
The device the module is on. Use it to keep your code device agnostic.
.. code-block:: python
def training_step(self, batch, batch_idx):
z = torch.rand(2, 3, device=self.device)
global_rank
~~~~~~~~~~~
The ``global_rank`` is the index of the current process across all nodes and devices.
Lightning will perform some operations such as logging, weight checkpointing only when ``global_rank=0``. You
usually do not need to use this property, but it is useful to know how to access it if needed.
.. code-block:: python
def training_step(self, batch, batch_idx):
if self.global_rank == 0:
# do something only once across all the nodes
...
global_step
~~~~~~~~~~~
The number of optimizer steps taken (does not reset each epoch).
This includes multiple optimizers and TBPTT steps (if enabled).
.. code-block:: python
def training_step(self, batch, batch_idx):
self.logger.experiment.log_image(..., step=self.global_step)
hparams
~~~~~~~
The arguments passed through ``LightningModule.__init__()`` and saved by calling
:meth:`~pytorch_lightning.core.mixins.hparams_mixin.HyperparametersMixin.save_hyperparameters` could be accessed by the ``hparams`` attribute.
.. code-block:: python
def __init__(self, learning_rate):
self.save_hyperparameters()
def configure_optimizers(self):
return Adam(self.parameters(), lr=self.hparams.learning_rate)
logger
~~~~~~
The current logger being used (tensorboard or other supported logger)
.. code-block:: python
def training_step(self, batch, batch_idx):
# the generic logger (same no matter if tensorboard or other supported logger)
self.logger
# the particular logger
tensorboard_logger = self.logger.experiment
loggers
~~~~~~~
The list of loggers currently being used by the Trainer.
.. code-block:: python
def training_step(self, batch, batch_idx):
# List of Logger objects
loggers = self.loggers
for logger in loggers:
logger.log_metrics({"foo": 1.0})
local_rank
~~~~~~~~~~~
The ``local_rank`` is the index of the current process across all the devices for the current node.
You usually do not need to use this property, but it is useful to know how to access it if needed.
For example, if using 10 machines (or nodes), the GPU at index 0 on each machine has local_rank = 0.
.. code-block:: python
def training_step(self, batch, batch_idx):
if self.local_rank == 0:
# do something only once across each node
...
precision
~~~~~~~~~
The type of precision used:
.. code-block:: python
def training_step(self, batch, batch_idx):
if self.precision == 16:
...
trainer
~~~~~~~
Pointer to the trainer
.. code-block:: python
def training_step(self, batch, batch_idx):
max_steps = self.trainer.max_steps
any_flag = self.trainer.any_flag
prepare_data_per_node
~~~~~~~~~~~~~~~~~~~~~
If set to ``True`` will call ``prepare_data()`` on LOCAL_RANK=0 for every node.
If set to ``False`` will only call from NODE_RANK=0, LOCAL_RANK=0.
.. testcode::
class LitModel(LightningModule):
def __init__(self):
super().__init__()
self.prepare_data_per_node = True
automatic_optimization
~~~~~~~~~~~~~~~~~~~~~~
When set to ``False``, Lightning does not automate the optimization process. This means you are responsible for handling
your optimizers. However, we do take care of precision and any accelerators used.
See :ref:`manual optimization ` for details.
.. code-block:: python
def __init__(self):
self.automatic_optimization = False
def training_step(self, batch, batch_idx):
opt = self.optimizers(use_pl_optimizer=True)
loss = ...
opt.zero_grad()
self.manual_backward(loss)
opt.step()
This is recommended only if using 2+ optimizers AND if you know how to perform the optimization procedure properly. Note
that automatic optimization can still be used with multiple optimizers by relying on the ``optimizer_idx`` parameter.
Manual optimization is most useful for research topics like reinforcement learning, sparse coding, and GAN research.
.. code-block:: python
def __init__(self):
self.automatic_optimization = False
def training_step(self, batch, batch_idx):
# access your optimizers with use_pl_optimizer=False. Default is True
opt_a, opt_b = self.optimizers(use_pl_optimizer=True)
gen_loss = ...
opt_a.zero_grad()
self.manual_backward(gen_loss)
opt_a.step()
disc_loss = ...
opt_b.zero_grad()
self.manual_backward(disc_loss)
opt_b.step()
example_input_array
~~~~~~~~~~~~~~~~~~~
Set and access example_input_array, which basically represents a single batch.
.. code-block:: python
def __init__(self):
self.example_input_array = ...
self.generator = ...
def on_train_epoch_end(self):
# generate some images using the example_input_array
gen_images = self.generator(self.example_input_array)
truncated_bptt_steps
~~~~~~~~~~~~~~~~~~~~
Truncated Backpropagation Through Time (TBPTT) performs perform backpropogation every k steps of
a much longer sequence. This is made possible by passing training batches
split along the time-dimensions into splits of size k to the
``training_step``. In order to keep the same forward propagation behavior, all
hidden states should be kept in-between each time-dimension split.
If this is enabled, your batches will automatically get truncated
and the Trainer will apply Truncated Backprop to it.
(`Williams et al. "An efficient gradient-based algorithm for on-line training of
recurrent network trajectories."
`_)
`Tutorial `_
.. testcode:: python
from pytorch_lightning import LightningModule
class MyModel(LightningModule):
def __init__(self, input_size, hidden_size, num_layers):
super().__init__()
# batch_first has to be set to True
self.lstm = nn.LSTM(
input_size=input_size,
hidden_size=hidden_size,
num_layers=num_layers,
batch_first=True,
)
...
# Important: This property activates truncated backpropagation through time
# Setting this value to 2 splits the batch into sequences of size 2
self.truncated_bptt_steps = 2
# Truncated back-propagation through time
def training_step(self, batch, batch_idx, hiddens):
x, y = batch
# the training step must be updated to accept a ``hiddens`` argument
# hiddens are the hiddens from the previous truncated backprop step
out, hiddens = self.lstm(x, hiddens)
...
return {"loss": ..., "hiddens": hiddens}
Lightning takes care of splitting your batch along the time-dimension. It is
assumed to be the second dimension of your batches. Therefore, in the
example above, we have set ``batch_first=True``.
.. code-block:: python
# we use the second as the time dimension
# (batch, time, ...)
sub_batch = batch[0, 0:t, ...]
To modify how the batch is split,
override the :meth:`pytorch_lightning.core.module.LightningModule.tbptt_split_batch` method:
.. testcode:: python
class LitMNIST(LightningModule):
def tbptt_split_batch(self, batch, split_size):
# do your own splitting on the batch
return splits
--------------
.. _lightning_hooks:
Hooks
=====
This is the pseudocode to describe the structure of :meth:`~pytorch_lightning.trainer.Trainer.fit`.
The inputs and outputs of each function are not represented for simplicity. Please check each function's API reference
for more information.
.. code-block:: python
def fit(self):
if global_rank == 0:
# prepare data is called on GLOBAL_ZERO only
prepare_data()
configure_callbacks()
with parallel(devices):
# devices can be GPUs, TPUs, ...
train_on_device(model)
def train_on_device(model):
# called PER DEVICE
setup("fit")
configure_optimizers()
on_fit_start()
# the sanity check runs here
on_train_start()
for epoch in epochs:
fit_loop()
on_train_end()
on_fit_end()
teardown("fit")
def fit_loop():
on_train_epoch_start()
for batch in train_dataloader():
on_train_batch_start()
on_before_batch_transfer()
transfer_batch_to_device()
on_after_batch_transfer()
training_step()
on_before_zero_grad()
optimizer_zero_grad()
on_before_backward()
backward()
on_after_backward()
on_before_optimizer_step()
configure_gradient_clipping()
optimizer_step()
on_train_batch_end()
if should_check_val:
val_loop()
# end training epoch
training_epoch_end()
on_train_epoch_end()
def val_loop():
on_validation_model_eval() # calls `model.eval()`
torch.set_grad_enabled(False)
on_validation_start()
on_validation_epoch_start()
val_outs = []
for batch_idx, batch in enumerate(val_dataloader()):
on_validation_batch_start(batch, batch_idx)
batch = on_before_batch_transfer(batch)
batch = transfer_batch_to_device(batch)
batch = on_after_batch_transfer(batch)
out = validation_step(batch, batch_idx)
on_validation_batch_end(batch, batch_idx)
val_outs.append(out)
validation_epoch_end(val_outs)
on_validation_epoch_end()
on_validation_end()
# set up for train
on_validation_model_train() # calls `model.train()`
torch.set_grad_enabled(True)
backward
~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.backward
:noindex:
on_before_backward
~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_before_backward
:noindex:
on_after_backward
~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_after_backward
:noindex:
on_before_zero_grad
~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_before_zero_grad
:noindex:
on_fit_start
~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_fit_start
:noindex:
on_fit_end
~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_fit_end
:noindex:
on_load_checkpoint
~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_load_checkpoint
:noindex:
on_save_checkpoint
~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_save_checkpoint
:noindex:
load_from_checkpoint
~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.load_from_checkpoint
:noindex:
on_train_start
~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_train_start
:noindex:
on_train_end
~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_train_end
:noindex:
on_validation_start
~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_validation_start
:noindex:
on_validation_end
~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_validation_end
:noindex:
on_test_batch_start
~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_test_batch_start
:noindex:
on_test_batch_end
~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_test_batch_end
:noindex:
on_test_epoch_start
~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_test_epoch_start
:noindex:
on_test_epoch_end
~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_test_epoch_end
:noindex:
on_test_start
~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_test_start
:noindex:
on_test_end
~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_test_end
:noindex:
on_predict_batch_start
~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_predict_batch_start
:noindex:
on_predict_batch_end
~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_predict_batch_end
:noindex:
on_predict_epoch_start
~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_predict_epoch_start
:noindex:
on_predict_epoch_end
~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_predict_epoch_end
:noindex:
on_predict_start
~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_predict_start
:noindex:
on_predict_end
~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_predict_end
:noindex:
on_train_batch_start
~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_train_batch_start
:noindex:
on_train_batch_end
~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_train_batch_end
:noindex:
on_train_epoch_start
~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_train_epoch_start
:noindex:
on_train_epoch_end
~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_train_epoch_end
:noindex:
on_validation_batch_start
~~~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_validation_batch_start
:noindex:
on_validation_batch_end
~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_validation_batch_end
:noindex:
on_validation_epoch_start
~~~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_validation_epoch_start
:noindex:
on_validation_epoch_end
~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_validation_epoch_end
:noindex:
configure_sharded_model
~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.configure_sharded_model
:noindex:
on_validation_model_eval
~~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_validation_model_eval
:noindex:
on_validation_model_train
~~~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_validation_model_train
:noindex:
on_test_model_eval
~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_test_model_eval
:noindex:
on_test_model_train
~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_test_model_train
:noindex:
on_before_optimizer_step
~~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_before_optimizer_step
:noindex:
configure_gradient_clipping
~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.configure_gradient_clipping
:noindex:
optimizer_step
~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.optimizer_step
:noindex:
optimizer_zero_grad
~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.optimizer_zero_grad
:noindex:
prepare_data
~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.prepare_data
:noindex:
setup
~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.setup
:noindex:
tbptt_split_batch
~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.tbptt_split_batch
:noindex:
teardown
~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.teardown
:noindex:
train_dataloader
~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.train_dataloader
:noindex:
val_dataloader
~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.val_dataloader
:noindex:
test_dataloader
~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.test_dataloader
:noindex:
predict_dataloader
~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.predict_dataloader
:noindex:
transfer_batch_to_device
~~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.transfer_batch_to_device
:noindex:
on_before_batch_transfer
~~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_before_batch_transfer
:noindex:
on_after_batch_transfer
~~~~~~~~~~~~~~~~~~~~~~~
.. automethod:: pytorch_lightning.core.module.LightningModule.on_after_batch_transfer
:noindex: