Rename loops/base.py to loops/loop.py

CHANGELOG.md

@@ -96,9 +96,6 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 - Raise an error if there are insufficient training batches when using a float value of `limit_train_batches` ([#12885](https://github.com/PyTorchLightning/pytorch-lightning/pull/12885))
 
-
-- Changed `pytorch_lightning.core.lightning` to `pytorch_lightning.core.module` ([#12740](https://github.com/PyTorchLightning/pytorch-lightning/pull/12740))
-
 ### Deprecated
 
 - Deprecated `pytorch_lightning.loggers.base.LightningLoggerBase` in favor of `pytorch_lightning.loggers.logger.Logger`, and deprecated `pytorch_lightning.loggers.base` in favor of `pytorch_lightning.loggers.logger` ([#120148](https://github.com/PyTorchLightning/pytorch-lightning/pull/12014))
@@ -116,6 +113,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Deprecated `pytorch_lightning.core.lightning.LightningModule` in favor of `pytorch_lightning.core.module.LightningModule` ([#12740](https://github.com/PyTorchLightning/pytorch-lightning/pull/12740))
 
 
+- Deprecated `pytorch_lightning.loops.base.Loop` in favor of `pytorch_lightning.loops.loop.Loop` ([#13043](https://github.com/PyTorchLightning/pytorch-lightning/pull/13043))
+
+
 - Deprecated `Trainer.reset_train_val_dataloaders()` in favor of `Trainer.reset_{train,val}_dataloader` ([#12184](https://github.com/PyTorchLightning/pytorch-lightning/pull/12184))
 
 ### Removed

docs/source/extensions/loops.rst

@@ -81,7 +81,7 @@ The core research logic is simply shifted to the :class:`~pytorch_lightning.core
         loss.backward()
         optimizer.step()
 
-Under the hood, the above loop is implemented using the :class:`~pytorch_lightning.loops.base.Loop` API like so:
+Under the hood, the above loop is implemented using the :class:`~pytorch_lightning.loops.loop.Loop` API like so:
 
 .. code-block:: python
 
@@ -183,7 +183,7 @@ Now your code is FULLY flexible and you can still leverage ALL the best parts of
 Creating a New Loop From Scratch
 --------------------------------
 
-You can also go wild and implement a full loop from scratch by sub-classing the :class:`~pytorch_lightning.loops.base.Loop` base class.
+You can also go wild and implement a full loop from scratch by sub-classing the :class:`~pytorch_lightning.loops.loop.Loop` base class.
 You will need to override a minimum of two things:
 
 .. code-block:: python
@@ -222,7 +222,7 @@ Loop API
 --------
 Here is the full API of methods available in the Loop base class.
 
-The :class:`~pytorch_lightning.loops.base.Loop` class is the base of all loops in the same way as the :class:`~pytorch_lightning.core.module.LightningModule` is the base of all models.
+The :class:`~pytorch_lightning.loops.loop.Loop` class is the base of all loops in the same way as the :class:`~pytorch_lightning.core.module.LightningModule` is the base of all models.
 It defines a public interface that each loop implementation must follow, the key ones are:
 
 Properties
@@ -231,13 +231,13 @@ Properties
 done
 ~~~~
 
-.. autoattribute:: pytorch_lightning.loops.base.Loop.done
+.. autoattribute:: pytorch_lightning.loops.loop.Loop.done
     :noindex:
 
 skip (optional)
 ~~~~~~~~~~~~~~~
 
-.. autoattribute:: pytorch_lightning.loops.base.Loop.skip
+.. autoattribute:: pytorch_lightning.loops.loop.Loop.skip
     :noindex:
 
 Methods
@@ -246,19 +246,19 @@ Methods
 reset (optional)
 ~~~~~~~~~~~~~~~~
 
-.. automethod:: pytorch_lightning.loops.base.Loop.reset
+.. automethod:: pytorch_lightning.loops.loop.Loop.reset
     :noindex:
 
 advance
 ~~~~~~~
 
-.. automethod:: pytorch_lightning.loops.base.Loop.advance
+.. automethod:: pytorch_lightning.loops.loop.Loop.advance
     :noindex:
 
 run (optional)
 ~~~~~~~~~~~~~~
 
-.. automethod:: pytorch_lightning.loops.base.Loop.run
+.. automethod:: pytorch_lightning.loops.loop.Loop.run
     :noindex:
 
 
@@ -267,7 +267,7 @@ run (optional)
 Subloops
 --------
 
-When you want to customize nested loops within loops, use the :meth:`~pytorch_lightning.loops.base.Loop.replace` method:
+When you want to customize nested loops within loops, use the :meth:`~pytorch_lightning.loops.loop.Loop.replace` method:
 
 .. code-block:: python
 
@@ -276,7 +276,7 @@ When you want to customize nested loops within loops, use the :meth:`~pytorch_li
     # Trainer runs the fit loop with your new epoch loop!
     trainer.fit(model)
 
-Alternatively, for more fine-grained control, use the :meth:`~pytorch_lightning.loops.base.Loop.connect` method:
+Alternatively, for more fine-grained control, use the :meth:`~pytorch_lightning.loops.loop.Loop.connect` method:
 
 .. code-block:: python
 
@@ -326,7 +326,7 @@ Here is what the structure would look like in plain Python:
                 ...
 
 
-Each of these :code:`for`-loops represents a class implementing the :class:`~pytorch_lightning.loops.base.Loop` interface.
+Each of these :code:`for`-loops represents a class implementing the :class:`~pytorch_lightning.loops.loop.Loop` interface.
 
 
 .. list-table:: Trainer entry points and associated loops

docs/source/extensions/loops_advanced.rst

@@ -18,7 +18,7 @@ A powerful property of the class-based loop interface is that it can own an inte
 Loop instances can save their state to the checkpoint through corresponding hooks and if implemented accordingly, resume the state of execution at the appropriate place.
 This design is particularly interesting for fault-tolerant training which is an experimental feature released in Lightning v1.5.
 
-The two hooks :meth:`~pytorch_lightning.loops.base.Loop.on_save_checkpoint` and :meth:`~pytorch_lightning.loops.base.Loop.on_load_checkpoint` function very similarly to how LightningModules and Callbacks save and load state.
+The two hooks :meth:`~pytorch_lightning.loops.loop.Loop.on_save_checkpoint` and :meth:`~pytorch_lightning.loops.loop.Loop.on_load_checkpoint` function very similarly to how LightningModules and Callbacks save and load state.
 
 .. code-block:: python
 
@@ -30,9 +30,9 @@ The two hooks :meth:`~pytorch_lightning.loops.base.Loop.on_save_checkpoint` and
     def on_load_checkpoint(self, state_dict):
         self.iteration = state_dict["iteration"]
 
-When the Trainer is restarting from a checkpoint (e.g., through :code:`trainer.fit(ckpt_path=...)`), the loop exposes a boolean attribute :attr:`~pytorch_lightning.loops.base.Loop.restarting`.
+When the Trainer is restarting from a checkpoint (e.g., through :code:`trainer.fit(ckpt_path=...)`), the loop exposes a boolean attribute :attr:`~pytorch_lightning.loops.loop.Loop.restarting`.
 Based around the value of this variable, the user can write the loop in such a way that it can restart from an arbitrary point given the state loaded from the checkpoint.
-For example, the implementation of the :meth:`~pytorch_lightning.loops.base.Loop.reset` method could look like this given our previous example:
+For example, the implementation of the :meth:`~pytorch_lightning.loops.loop.Loop.reset` method could look like this given our previous example:
 
 .. code-block:: python
 

pl_examples/loop_examples/kfold.py

@@ -31,8 +31,8 @@
 from pl_examples.basic_examples.mnist_examples.image_classifier_4_lightning_module import ImageClassifier
 from pytorch_lightning import LightningDataModule, seed_everything, Trainer
 from pytorch_lightning.core.module import LightningModule
-from pytorch_lightning.loops.base import Loop
 from pytorch_lightning.loops.fit_loop import FitLoop
+from pytorch_lightning.loops.loop import Loop
 from pytorch_lightning.trainer.states import TrainerFn
 
 #############################################################################################

pytorch_lightning/loops/__init__.py

@@ -11,11 +11,9 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-
-from pytorch_lightning.loops.base import Loop  # noqa: F401
-from pytorch_lightning.loops.batch import ManualOptimization  # noqa: F401
+from pytorch_lightning.loops.loop import Loop  # noqa: F401 isort: skip (avoids circular imports)
 from pytorch_lightning.loops.batch import TrainingBatchLoop  # noqa: F401
 from pytorch_lightning.loops.dataloader import DataLoaderLoop, EvaluationLoop, PredictionLoop  # noqa: F401
 from pytorch_lightning.loops.epoch import EvaluationEpochLoop, PredictionEpochLoop, TrainingEpochLoop  # noqa: F401
 from pytorch_lightning.loops.fit_loop import FitLoop  # noqa: F401
-from pytorch_lightning.loops.optimization.optimizer_loop import OptimizerLoop  # noqa: F401
+from pytorch_lightning.loops.optimization import ManualOptimization, OptimizerLoop  # noqa: F401