Rooba
/
lightning
mirror of https://github.com/Lightning-AI/lightning.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Consistent imports in docs for core APIs (#18869) 

Co-authored-by: Sebastian Raschka <mail@sebastianraschka.com>
This commit is contained in:
Adrian Wälchli 2023-10-27 20:41:29 +02:00 committed by GitHub
parent c1437ccadf
commit f6a36cf220
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
23 changed files with 103 additions and 93 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
docs/source-pytorch/accelerators/tpu_faq.rst
View File

@ -88,10 +88,10 @@ How to setup the debug mode for Training on TPUs?
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
my_model = MyLightningModule()
trainer = pl.Trainer(accelerator="tpu", devices=8, strategy="xla_debug")
trainer = L.Trainer(accelerator="tpu", devices=8, strategy="xla_debug")
trainer.fit(my_model)
Example Metrics report:

8
docs/source-pytorch/accelerators/tpu_intermediate.rst
View File

@ -44,10 +44,10 @@ To use a full TPU pod skip to the TPU pod section.
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
my_model = MyLightningModule()
trainer = pl.Trainer(accelerator="tpu", devices=8)
trainer = L.Trainer(accelerator="tpu", devices=8)
trainer.fit(my_model)
That's it! Your model will train on all 8 TPU cores.
@ -113,10 +113,10 @@ By default, TPU training will use 32-bit precision. To enable it, do
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
my_model = MyLightningModule()
trainer = pl.Trainer(accelerator="tpu", precision="16-true")
trainer = L.Trainer(accelerator="tpu", precision="16-true")
trainer.fit(my_model)
Under the hood the xla library will use the `bfloat16 type <https://en.wikipedia.org/wiki/Bfloat16_floating-point_format>`_.

13
docs/source-pytorch/advanced/model_parallel/deepspeed.rst
View File

@ -132,12 +132,11 @@ For even more speed benefit, DeepSpeed offers an optimized CPU version of ADAM c
.. code-block:: python
import lightning.pytorch
from lightning.pytorch import Trainer
from lightning.pytorch import LightningModule, Trainer
from deepspeed.ops.adam import DeepSpeedCPUAdam
class MyModel(pl.LightningModule):
class MyModel(LightningModule):
...
def configure_optimizers(self):
@ -180,7 +179,7 @@ Also please have a look at our :ref:`deepspeed-zero-stage-3-tips` which contains
from deepspeed.ops.adam import FusedAdam
class MyModel(pl.LightningModule):
class MyModel(LightningModule):
...
def configure_optimizers(self):
@ -202,7 +201,7 @@ You can also use the Lightning Trainer to run predict or evaluate with DeepSpeed
from lightning.pytorch import Trainer
class MyModel(pl.LightningModule):
class MyModel(LightningModule):
...
@ -228,7 +227,7 @@ This reduces the time taken to initialize very large models, as well as ensure w
from deepspeed.ops.adam import FusedAdam
class MyModel(pl.LightningModule):
class MyModel(LightningModule):
...
def configure_model(self):
@ -367,7 +366,7 @@ This saves memory when training larger models, however requires using a checkpoi
import deepspeed
class MyModel(pl.LightningModule):
class MyModel(LightningModule):
...
def configure_model(self):

4
docs/source-pytorch/advanced/training_tricks.rst
View File

@ -398,7 +398,7 @@ The :class:`~lightning.pytorch.core.datamodule.LightningDataModule` class provid
.. code-block:: python
class MNISTDataModule(pl.LightningDataModule):
class MNISTDataModule(L.LightningDataModule):
def prepare_data(self):
MNIST(self.data_dir, download=True)
@ -421,7 +421,7 @@ For this, all data pre-loading should be done on the main process inside :meth:`
.. code-block:: python
class MNISTDataModule(pl.LightningDataModule):
class MNISTDataModule(L.LightningDataModule):
def __init__(self, data_dir: str):
self.mnist = MNIST(data_dir, download=True, transform=T.ToTensor())

2
docs/source-pytorch/cli/lightning_cli_advanced.rst
View File

@ -164,7 +164,7 @@ to the class constructor. For example, your model is defined as:
.. code:: python
# model.py
class MyModel(pl.LightningModule):
class MyModel(L.LightningModule):
def __init__(self, criterion: torch.nn.Module):
self.criterion = criterion

11
docs/source-pytorch/common/checkpointing_advanced.rst
View File

@ -54,9 +54,9 @@ Modify a checkpoint anywhere
****************************
When you need to change the components of a checkpoint before saving or loading, use the :meth:`~lightning.pytorch.core.hooks.CheckpointHooks.on_save_checkpoint` and :meth:`~lightning.pytorch.core.hooks.CheckpointHooks.on_load_checkpoint` of your ``LightningModule``.
.. code:: python
.. code-block:: python
class LitModel(pl.LightningModule):
class LitModel(L.LightningModule):
def on_save_checkpoint(self, checkpoint):
checkpoint["something_cool_i_want_to_save"] = my_cool_pickable_object
@ -65,9 +65,12 @@ When you need to change the components of a checkpoint before saving or loading,
Use the above approach when you need to couple this behavior to your LightningModule for reproducibility reasons. Otherwise, Callbacks also have the :meth:`~lightning.pytorch.callbacks.callback.Callback.on_save_checkpoint` and :meth:`~lightning.pytorch.callbacks.callback.Callback.on_load_checkpoint` which you should use instead:
.. code:: python
.. code-block:: python
class LitCallback(pl.Callback):
import lightning as L
class LitCallback(L.Callback):
def on_save_checkpoint(self, checkpoint):
checkpoint["something_cool_i_want_to_save"] = my_cool_pickable_object

4
docs/source-pytorch/common/checkpointing_basic.rst
View File

@ -127,7 +127,7 @@ In some cases, we may also pass entire PyTorch modules to the ``__init__`` metho
.. code-block:: python
class LitAutoencoder(pl.LightningModule):
class LitAutoencoder(L.LightningModule):
def __init__(self, encoder, decoder):
...
@ -160,7 +160,7 @@ For example, let's pretend we created a LightningModule like so:
...
class Autoencoder(pl.LightningModule):
class Autoencoder(L.LightningModule):
def __init__(self, encoder, decoder, *args, **kwargs):
...

2
docs/source-pytorch/common/checkpointing_intermediate.rst
View File

@ -27,7 +27,7 @@ Any value that has been logged via *self.log* in the LightningModule can be moni
.. code-block:: python
class LitModel(pl.LightningModule):
class LitModel(L.LightningModule):
def training_step(self, batch, batch_idx):
self.log("my_metric", x)

6
docs/source-pytorch/common/evaluation_basic.rst
View File

@ -39,7 +39,7 @@ To add a test loop, implement the **test_step** method of the LightningModule
.. code:: python
class LitAutoEncoder(pl.LightningModule):
class LitAutoEncoder(L.LightningModule):
def training_step(self, batch, batch_idx):
...
@ -99,7 +99,7 @@ To add a validation loop, implement the **validation_step** method of the Lightn
.. code:: python
class LitAutoEncoder(pl.LightningModule):
class LitAutoEncoder(L.LightningModule):
def training_step(self, batch, batch_idx):
...
@ -127,5 +127,5 @@ To run the validation loop, pass in the validation set to **.fit**
model = LitAutoEncoder(...)
# train with both splits
trainer = pl.Trainer()
trainer = L.Trainer()
trainer.fit(model, train_loader, valid_loader)

2
docs/source-pytorch/common/evaluation_intermediate.rst
View File

@ -121,7 +121,7 @@ you can also pass in an :doc:`datamodules <../data/datamodule>` that have overri
.. code-block:: python
class MyDataModule(pl.LightningDataModule):
class MyDataModule(L.LightningDataModule):
...
def test_dataloader(self):

42
docs/source-pytorch/common/lightning_module.rst
View File

@ -84,13 +84,13 @@ Here are the only required methods.
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
import torch
from lightning.pytorch.demos import Transformer
class LightningTransformer(pl.LightningModule):
class LightningTransformer(L.LightningModule):
def __init__(self, vocab_size):
super().__init__()
self.model = Transformer(vocab_size=vocab_size)
@ -118,7 +118,7 @@ Which you can train by doing:
dataloader = DataLoader(dataset)
model = LightningTransformer(vocab_size=dataset.vocab_size)
trainer = pl.Trainer(fast_dev_run=100)
trainer = L.Trainer(fast_dev_run=100)
trainer.fit(model=model, train_dataloaders=dataloader)
The LightningModule has many convenient methods, but the core ones you need to know about are:
@ -157,7 +157,7 @@ To activate the training loop, override the :meth:`~lightning.pytorch.core.Light
.. code-block:: python
class LightningTransformer(pl.LightningModule):
class LightningTransformer(L.LightningModule):
def __init__(self, vocab_size):
super().__init__()
self.model = Transformer(vocab_size=vocab_size)
@ -235,7 +235,7 @@ override the :meth:`~lightning.pytorch.LightningModule.on_train_epoch_end` metho
.. code-block:: python
class LightningTransformer(pl.LightningModule):
class LightningTransformer(L.LightningModule):
def __init__(self, vocab_size):
super().__init__()
self.model = Transformer(vocab_size=vocab_size)
@ -269,7 +269,7 @@ To activate the validation loop while training, override the :meth:`~lightning.p
.. code-block:: python
class LightningTransformer(pl.LightningModule):
class LightningTransformer(L.LightningModule):
def validation_step(self, batch, batch_idx):
inputs, target = batch
output = self.model(inputs, target)
@ -306,7 +306,7 @@ and calling :meth:`~lightning.pytorch.trainer.trainer.Trainer.validate`.
.. code-block:: python
model = LightningTransformer(vocab_size=dataset.vocab_size)
trainer = pl.Trainer()
trainer = L.Trainer()
trainer.validate(model)
.. note::
@ -327,7 +327,7 @@ Note that this method is called before :meth:`~lightning.pytorch.LightningModule
.. code-block:: python
class LightningTransformer(pl.LightningModule):
class LightningTransformer(L.LightningModule):
def __init__(self, vocab_size):
super().__init__()
self.model = Transformer(vocab_size=vocab_size)
@ -366,7 +366,7 @@ The only difference is that the test loop is only called when :meth:`~lightning.
model = LightningTransformer(vocab_size=dataset.vocab_size)
dataloader = DataLoader(dataset)
trainer = pl.Trainer()
trainer = L.Trainer()
trainer.fit(model=model, train_dataloaders=dataloader)
# automatically loads the best weights for you
@ -377,7 +377,7 @@ There are two ways to call ``test()``:
.. code-block:: python
# call after training
trainer = pl.Trainer()
trainer = L.Trainer()
trainer.fit(model=model, train_dataloaders=dataloader)
# automatically auto-loads the best weights from the previous run
@ -387,7 +387,7 @@ There are two ways to call ``test()``:
model = LightningTransformer.load_from_checkpoint(PATH)
dataset = WikiText2()
test_dataloader = DataLoader(dataset)
trainer = pl.Trainer()
trainer = L.Trainer()
trainer.test(model, dataloaders=test_dataloader)
.. note::
@ -420,7 +420,7 @@ For the example let's override ``predict_step``:
.. code-block:: python
class LightningTransformer(pl.LightningModule):
class LightningTransformer(L.LightningModule):
def __init__(self, vocab_size):
super().__init__()
self.model = Transformer(vocab_size=vocab_size)
@ -447,7 +447,7 @@ There are two ways to call ``predict()``:
.. code-block:: python
# call after training
trainer = pl.Trainer()
trainer = L.Trainer()
trainer.fit(model=model, train_dataloaders=dataloader)
# automatically auto-loads the best weights from the previous run
@ -457,7 +457,7 @@ There are two ways to call ``predict()``:
model = LightningTransformer.load_from_checkpoint(PATH)
dataset = WikiText2()
test_dataloader = DataLoader(dataset)
trainer = pl.Trainer()
trainer = L.Trainer()
predictions = trainer.predict(model, dataloaders=test_dataloader)
Inference in Research
@ -469,7 +469,7 @@ If you want to perform inference with the system, you can add a ``forward`` meth
.. code-block:: python
class LightningTransformer(pl.LightningModule):
class LightningTransformer(L.LightningModule):
def __init__(self, vocab_size):
super().__init__()
self.model = Transformer(vocab_size=vocab_size)
@ -500,7 +500,7 @@ such as text generation:
.. code-block:: python
class Seq2Seq(pl.LightningModule):
class Seq2Seq(L.LightningModule):
def forward(self, x):
embeddings = self(x)
hidden_states = self.encoder(embeddings)
@ -514,7 +514,7 @@ In the case where you want to scale your inference, you should be using
.. code-block:: python
class Autoencoder(pl.LightningModule):
class Autoencoder(L.LightningModule):
def forward(self, x):
return self.decoder(x)
@ -538,7 +538,7 @@ For cases like production, you might want to iterate different models inside a L
from torchmetrics.functional import accuracy
class ClassificationTask(pl.LightningModule):
class ClassificationTask(L.LightningModule):
def __init__(self, model):
super().__init__()
self.model = model
@ -590,7 +590,7 @@ Tasks can be arbitrarily complex such as implementing GAN training, self-supervi
.. code-block:: python
class GANTask(pl.LightningModule):
class GANTask(L.LightningModule):
def __init__(self, generator, discriminator):
super().__init__()
self.generator = generator
@ -643,7 +643,7 @@ checkpoint, which simplifies model re-instantiation after training.
.. code-block:: python
class LitMNIST(pl.LightningModule):
class LitMNIST(L.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
@ -667,7 +667,7 @@ parameters should be provided back when reloading the LightningModule. In this c
.. code-block:: python
class LitMNIST(pl.LightningModule):
class LitMNIST(L.LightningModule):
def __init__(self, loss_fx, generator_network, layer_1_dim=128):
super().__init__()
self.layer_1_dim = layer_1_dim

36
docs/source-pytorch/data/datamodule.rst
View File

@ -79,7 +79,7 @@ The equivalent DataModule just organizes the same exact code, but makes it reusa
.. code-block:: python
class MNISTDataModule(pl.LightningDataModule):
class MNISTDataModule(L.LightningDataModule):
def __init__(self, data_dir: str = "path/to/dir", batch_size: int = 32):
super().__init__()
self.data_dir = data_dir
@ -125,7 +125,7 @@ Here's a more realistic, complex DataModule that shows how much more reusable th
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
from torch.utils.data import random_split, DataLoader
# Note - you must have torchvision installed for this example
@ -133,7 +133,7 @@ Here's a more realistic, complex DataModule that shows how much more reusable th
from torchvision import transforms
class MNISTDataModule(pl.LightningDataModule):
class MNISTDataModule(L.LightningDataModule):
def __init__(self, data_dir: str = "./"):
super().__init__()
self.data_dir = data_dir
@ -171,8 +171,10 @@ Here's a more realistic, complex DataModule that shows how much more reusable th
def predict_dataloader(self):
return DataLoader(self.mnist_predict, batch_size=32)
---------------
***********************
LightningDataModule API
***********************
@ -200,7 +202,7 @@ depends upon :ref:`prepare_data_per_node<data/datamodule:prepare_data_per_node>`
.. code-block:: python
class MNISTDataModule(pl.LightningDataModule):
class MNISTDataModule(L.LightningDataModule):
def prepare_data(self):
# download
MNIST(os.getcwd(), train=True, download=True, transform=transforms.ToTensor())
@ -226,10 +228,10 @@ There are also data operations you might want to perform on every GPU. Use :meth
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
class MNISTDataModule(pl.LightningDataModule):
class MNISTDataModule(L.LightningDataModule):
def setup(self, stage: str):
# Assign Train/val split(s) for use in Dataloaders
if stage == "fit":
@ -247,7 +249,7 @@ For eg., if you are working with NLP task where you need to tokenize the text an
.. code-block:: python
class LitDataModule(LightningDataModule):
class LitDataModule(L.LightningDataModule):
def prepare_data(self):
dataset = load_Dataset(...)
train_dataset = ...
@ -275,10 +277,10 @@ Usually you just wrap the dataset you defined in :ref:`setup<data/datamodule:set
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
class MNISTDataModule(pl.LightningDataModule):
class MNISTDataModule(L.LightningDataModule):
def train_dataloader(self):
return DataLoader(self.mnist_train, batch_size=64)
@ -292,10 +294,10 @@ Usually you just wrap the dataset you defined in :ref:`setup<data/datamodule:set
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
class MNISTDataModule(pl.LightningDataModule):
class MNISTDataModule(L.LightningDataModule):
def val_dataloader(self):
return DataLoader(self.mnist_val, batch_size=64)
@ -310,10 +312,10 @@ Usually you just wrap the dataset you defined in :ref:`setup<data/datamodule:set
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
class MNISTDataModule(pl.LightningDataModule):
class MNISTDataModule(L.LightningDataModule):
def test_dataloader(self):
return DataLoader(self.mnist_test, batch_size=64)
@ -326,10 +328,10 @@ Usually you just wrap the dataset you defined in :ref:`setup<data/datamodule:set
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
class MNISTDataModule(pl.LightningDataModule):
class MNISTDataModule(L.LightningDataModule):
def predict_dataloader(self):
return DataLoader(self.mnist_predict, batch_size=64)
@ -468,10 +470,10 @@ Like LightningModules, DataModules support hyperparameters with the same API.
.. code-block:: python
import lightning.pytorch as pl
import lightning as L
class CustomDataModule(pl.LightningDataModule):
class CustomDataModule(L.LightningDataModule):
def __init__(self, *args, **kwargs):
super().__init__()
self.save_hyperparameters()

2
docs/source-pytorch/deploy/production_advanced_2.rst
View File

@ -47,7 +47,7 @@ If you want to script a different method, you can decorate the method with :func
.. code-block:: python
class LitMCdropoutModel(pl.LightningModule):
class LitMCdropoutModel(L.LightningModule):
def __init__(self, model, mc_iteration):
super().__init__()
self.model = model

2
docs/source-pytorch/deploy/production_basic.rst
View File

@ -50,7 +50,7 @@ When you need to add complicated pre-processing or post-processing logic to your
.. code-block:: python
class LitMCdropoutModel(pl.LightningModule):
class LitMCdropoutModel(L.LightningModule):
def __init__(self, model, mc_iteration):
super().__init__()
self.model = model

6
docs/source-pytorch/ecosystem/asr_nlp_tts.rst
View File

@ -395,7 +395,7 @@ Developing NER Model From Scratch
# hydra_runner calls hydra.main and is useful for multi-node experiments
@hydra_runner(config_path="conf", config_name="token_classification_config")
def main(cfg: DictConfig) -> None:
trainer = pl.Trainer(**cfg.trainer)
trainer = L.Trainer(**cfg.trainer)
model = TokenClassificationModel(cfg.model, trainer=trainer)
trainer.fit(model)
@ -406,7 +406,7 @@ Inference from file:
.. code-block:: python
gpu = 1 if cfg.trainer.gpus != 0 else 0
trainer = pl.Trainer(accelerator="gpu", devices=gpu)
trainer = L.Trainer(accelerator="gpu", devices=gpu)
model.set_trainer(trainer)
model.evaluate_from_file(
text_file=os.path.join(cfg.model.dataset.data_dir, cfg.model.validation_ds.text_file),
@ -638,7 +638,7 @@ Developing TTS Model From Scratch
# hydra_runner calls hydra.main and is useful for multi-node experiments
@hydra_runner(config_path="conf", config_name="glow_tts")
def main(cfg):
trainer = pl.Trainer(**cfg.trainer)
trainer = L.Trainer(**cfg.trainer)
model = GlowTTSModel(cfg=cfg.model, trainer=trainer)
trainer.fit(model)

4
docs/source-pytorch/ecosystem/bolts.rst
View File

@ -83,9 +83,9 @@ We also have a collection of callbacks.
.. code-block:: python
from pl_bolts.callbacks import PrintTableMetricsCallback
import lightning.pytorch as pl
import lightning as L
trainer = pl.Trainer(callbacks=[PrintTableMetricsCallback()])
trainer = L.Trainer(callbacks=[PrintTableMetricsCallback()])
# loss│train_loss│val_loss│epoch
# ──────────────────────────────

5
docs/source-pytorch/extensions/datamodules_state.rst
View File

@ -4,7 +4,10 @@ When a checkpoint is created, it asks every DataModule for their state. If your
.. code:: python
class LitDataModule(pl.DataModuler):
import lightning as L
class LitDataModule(L.LightningDataModule):
def state_dict(self):
# track whatever you want here
state = {"current_train_batch_index": self.current_train_batch_index}

5
docs/source-pytorch/model/build_model_advanced.rst
View File

@ -16,7 +16,10 @@ Inject custom code anywhere in the Training loop using any of the 20+ methods (:
.. testcode::
class LitModel(pl.LightningModule):
import lightning as L
class LitModel(L.LightningModule):
def backward(self, loss):
loss.backward()

6
docs/source-pytorch/model/train_model_basic.rst
View File

@ -21,7 +21,7 @@ Add the relevant imports at the top of the file
from torchvision import transforms
from torchvision.datasets import MNIST
from torch.utils.data import DataLoader
import lightning.pytorch as pl
import lightning as L
----
@ -60,7 +60,7 @@ The LightningModule is the full **recipe** that defines how your nn.Modules inte
.. code:: python
class LitAutoEncoder(pl.LightningModule):
class LitAutoEncoder(L.LightningModule):
def __init__(self, encoder, decoder):
super().__init__()
self.encoder = encoder
@ -104,7 +104,7 @@ To train the model use the Lightning :doc:`Trainer <../common/trainer>` which ha
autoencoder = LitAutoEncoder(Encoder(), Decoder())
# train model
trainer = pl.Trainer()
trainer = L.Trainer()
trainer.fit(model=autoencoder, train_dataloaders=train_loader)
----

16
docs/source-pytorch/starter/converting.rst
View File

@ -16,7 +16,7 @@ Keep your regular nn.Module architecture
.. testcode::
import lightning.pytorch as pl
import lightning as L
import torch
import torch.nn as nn
import torch.nn.functional as F
@ -44,7 +44,7 @@ In the training_step of the LightningModule configure how your training routine
.. testcode::
class LitModel(pl.LightningModule):
class LitModel(L.LightningModule):
def __init__(self, encoder):
super().__init__()
self.encoder = encoder
@ -66,7 +66,7 @@ Move your optimizers to the :meth:`~lightning.pytorch.core.LightningModule.confi
.. testcode::
class LitModel(pl.LightningModule):
class LitModel(L.LightningModule):
def configure_optimizers(self):
optimizer = torch.optim.Adam(self.encoder.parameters(), lr=1e-3)
lr_scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=1)
@ -81,7 +81,7 @@ If you need a validation loop, configure how your validation routine behaves wit
.. testcode::
class LitModel(pl.LightningModule):
class LitModel(L.LightningModule):
def validation_step(self, batch, batch_idx):
x, y = batch
y_hat = self.encoder(x)
@ -99,7 +99,7 @@ If you need a test loop, configure how your testing routine behaves with a batch
.. testcode::
class LitModel(pl.LightningModule):
class LitModel(L.LightningModule):
def test_step(self, batch, batch_idx):
x, y = batch
y_hat = self.encoder(x)
@ -115,7 +115,7 @@ If you need a prediction loop, configure how your prediction routine behaves wit
.. testcode::
class LitModel(LightningModule):
class LitModel(L.LightningModule):
def predict_step(self, batch, batch_idx):
x, y = batch
pred = self.encoder(x)
@ -135,7 +135,7 @@ If you still need to access the current device, you can use ``self.device`` anyw
.. testcode::
class LitModel(LightningModule):
class LitModel(L.LightningModule):
def training_step(self, batch, batch_idx):
z = torch.randn(4, 5, device=self.device)
...
@ -145,7 +145,7 @@ Hint: If you are initializing a :class:`~torch.Tensor` within the ``LightningMod
.. testcode::
class LitModel(LightningModule):
class LitModel(L.LightningModule):
def __init__(self):
super().__init__()
self.register_buffer("running_mean", torch.zeros(num_features))

12
docs/source-pytorch/starter/introduction.rst
View File

@ -112,7 +112,7 @@ A LightningModule enables your PyTorch nn.Module to play together in complex way
from torch import optim, nn, utils, Tensor
from torchvision.datasets import MNIST
from torchvision.transforms import ToTensor
import lightning.pytorch as pl
import lightning as L
# define any number of nn.Modules (or use your current ones)
encoder = nn.Sequential(nn.Linear(28 * 28, 64), nn.ReLU(), nn.Linear(64, 3))
@ -120,7 +120,7 @@ A LightningModule enables your PyTorch nn.Module to play together in complex way
# define the LightningModule
class LitAutoEncoder(pl.LightningModule):
class LitAutoEncoder(L.LightningModule):
def __init__(self, encoder, decoder):
super().__init__()
self.encoder = encoder
@ -171,7 +171,7 @@ The Lightning :doc:`Trainer <../common/trainer>` "mixes" any :doc:`LightningModu
.. code-block:: python
# train the model (hint: here are some helpful Trainer arguments for rapid idea iteration)
trainer = pl.Trainer(limit_train_batches=100, max_epochs=1)
trainer = L.Trainer(limit_train_batches=100, max_epochs=1)
trainer.fit(model=autoencoder, train_dataloaders=train_loader)
The Lightning :doc:`Trainer <../common/trainer>` automates `40+ tricks <../common/trainer.html#trainer-flags>`_ including:
@ -237,7 +237,7 @@ Enable advanced training features using Trainer arguments. These are state-of-th
)
# train 1TB+ parameter models with Deepspeed/fsdp
trainer = Trainer(
trainer = L.Trainer(
devices=4,
accelerator="gpu",
strategy="deepspeed_stage_2",
@ -245,7 +245,7 @@ Enable advanced training features using Trainer arguments. These are state-of-th
)
# 20+ helpful flags for rapid idea iteration
trainer = Trainer(
trainer = L.Trainer(
max_epochs=10,
min_epochs=5,
overfit_batches=1
@ -276,7 +276,7 @@ Inject custom code anywhere in the Training loop using any of the 20+ methods (:
.. testcode::
class LitAutoEncoder(pl.LightningModule):
class LitAutoEncoder(L.LightningModule):
def backward(self, loss):
loss.backward()

2
docs/source-pytorch/starter/style_guide.rst
View File

@ -139,7 +139,7 @@ In practice, the code looks like this:
.. code-block::
class LitModel(pl.LightningModule):
class LitModel(L.LightningModule):
def __init__(...):

2
docs/source-pytorch/visualize/logging_basic.rst
View File

@ -27,7 +27,7 @@ To track a metric, simply use the *self.log* method available inside the *Lightn
.. code-block:: python
class LitModel(pl.LightningModule):
class LitModel(L.LightningModule):
def training_step(self, batch, batch_idx):
value = ...
self.log("some_value", value)