Enables reload of dataloaders on every n epochs from every epoch

CHANGELOG.md

@@ -212,6 +212,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 - MLflowLogger now uses the env variable `MLFLOW_TRACKING_URI` as default tracking uri ([#7457](https://github.com/PyTorchLightning/pytorch-lightning/pull/7457))
 
+- Changed `Trainer` arg and functionality from `reload_dataloaders_every_epoch` to `reload_dataloaders_every_n_epochs` ([#5043](https://github.com/PyTorchLightning/pytorch-lightning/pull/5043))
 
 - Changed `WandbLogger(log_model={True/'all'})` to log models as artifacts ([#6231](https://github.com/PyTorchLightning/pytorch-lightning/pull/6231))
 
@@ -288,6 +289,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Deprecated `mode` parameter in `ModelSummary` in favor of `max_depth` ([#8062](https://github.com/PyTorchLightning/pytorch-lightning/pull/8062))
 
 
+- Deprecated `reload_dataloaders_every_epoch` argument of `Trainer` in favor of `reload_dataloaders_every_n_epochs` ([#5043](https://github.com/PyTorchLightning/pytorch-lightning/pull/5043))
+
+
 ### Removed
 
 - Removed `ProfilerConnector` ([#7654](https://github.com/PyTorchLightning/pytorch-lightning/pull/7654))
@@ -708,6 +712,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
     [#6323](https://github.com/PyTorchLightning/pytorch-lightning/pull/6323),
     [#6211](https://github.com/PyTorchLightning/pytorch-lightning/pull/6211))
 
+
 ## [1.2.9] - 2021-04-20
 
 ### Fixed
@@ -752,8 +757,6 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Enforce an epoch scheduler interval when using SWA ([#6588](https://github.com/PyTorchLightning/pytorch-lightning/pull/6588))
 - Fixed TPU Colab hang issue, post training ([#6816](https://github.com/PyTorchLightning/pytorch-lightning/pull/6816))
 - Fixed a bug where `TensorBoardLogger` would give a warning and not log correctly to a symbolic link `save_dir` ([#6730](https://github.com/PyTorchLightning/pytorch-lightning/pull/6730))
-
-
 - Fixed bug where `predict` could not be used when `progress_bar_refresh_rate=0` ([#6884](https://github.com/PyTorchLightning/pytorch-lightning/pull/6884))
 
 

docs/source/common/trainer.rst

@@ -1297,8 +1297,8 @@ Note:
       Lightning will set it to 20 in these environments if the user does not provide a value.
     - This argument is ignored if a custom callback is passed to :paramref:`~Trainer.callbacks`.
 
-reload_dataloaders_every_epoch
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+reload_dataloaders_every_n_epochs
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 .. raw:: html
 
@@ -1308,19 +1308,20 @@ reload_dataloaders_every_epoch
 
 |
 
-Set to True to reload dataloaders every epoch.
+Set to a postive integer to reload dataloaders every n epochs.
 
 .. code-block:: python
 
-    # if False (default)
+    # if 0 (default)
     train_loader = model.train_dataloader()
     for epoch in epochs:
         for batch in train_loader:
             ...
 
-    # if True
+    # if a positive integer
     for epoch in epochs:
-        train_loader = model.train_dataloader()
+        if not epoch % reload_dataloaders_every_n_epochs:
+            train_loader = model.train_dataloader()
         for batch in train_loader:
 
 .. _replace-sampler-ddp:

pytorch_lightning/core/hooks.py

@@ -435,8 +435,9 @@ def train_dataloader(self) -> TRAIN_DATALOADERS:
             A collection of :class:`torch.utils.data.DataLoader` specifying training samples.
             In the case of multiple dataloaders, please see this :ref:`page <multiple-training-dataloaders>`.
 
-        The dataloader you return will not be called every epoch unless you set
-        :paramref:`~pytorch_lightning.trainer.Trainer.reload_dataloaders_every_epoch` to ``True``.
+        The dataloader you return will not be reloaded unless you set
+        :paramref:`~pytorch_lightning.trainer.Trainer.reload_dataloaders_every_n_epochs` to
+        a positive integer.
 
         For data processing use the following pattern:
 
@@ -505,8 +506,9 @@ def test_dataloader(self) -> EVAL_DATALOADERS:
         r"""
         Implement one or multiple PyTorch DataLoaders for testing.
 
-        The dataloader you return will not be called every epoch unless you set
-        :paramref:`~pytorch_lightning.trainer.Trainer.reload_dataloaders_every_epoch` to ``True``.
+        The dataloader you return will not be reloaded unless you set
+        :paramref:`~pytorch_lightning.trainer.Trainer.reload_dataloaders_every_n_epochs` to
+        a postive integer.
 
         For data processing use the following pattern:
 
@@ -565,8 +567,9 @@ def val_dataloader(self) -> EVAL_DATALOADERS:
         r"""
         Implement one or multiple PyTorch DataLoaders for validation.
 
-        The dataloader you return will not be called every epoch unless you set
-        :paramref:`~pytorch_lightning.trainer.Trainer.reload_dataloaders_every_epoch` to ``True``.
+        The dataloader you return will not be reloaded unless you set
+        :paramref:`~pytorch_lightning.trainer.Trainer.reload_dataloaders_every_n_epochs` to
+        a positive integer.
 
         It's recommended that all data downloads and preparation happen in :meth:`prepare_data`.
 

pytorch_lightning/loops/dataloader/evaluation_loop.py

@@ -180,7 +180,7 @@ def reload_evaluation_dataloaders(self) -> None:
         model = self.trainer.lightning_module
         if self.trainer.testing:
             self.trainer.reset_test_dataloader(model)
-        elif self.trainer.val_dataloaders is None or self.trainer.reload_dataloaders_every_epoch:
+        elif self.trainer.val_dataloaders is None or self.trainer._should_reload_dl_epoch:
             self.trainer.reset_val_dataloader(model)
 
     def on_evaluation_start(self, *args: Any, **kwargs: Any) -> None:

pytorch_lightning/loops/fit_loop.py

@@ -191,7 +191,7 @@ def on_advance_start(self) -> None:
         model = self.trainer.lightning_module
 
         # reset train dataloader
-        if self.current_epoch != 0 and self.trainer.reload_dataloaders_every_epoch:
+        if self.current_epoch != 0 and self.trainer._should_reload_dl_epoch:
             self.trainer.reset_train_dataloader(model)
 
         # TODO: specify the possible exception

pytorch_lightning/trainer/connectors/data_connector.py

@@ -16,6 +16,7 @@
 
 import pytorch_lightning as pl
 from pytorch_lightning.trainer.supporters import prefetch_iterator
+from pytorch_lightning.utilities import rank_zero_deprecation
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
 from pytorch_lightning.utilities.model_helpers import is_overridden
 from pytorch_lightning.utilities.types import EVAL_DATALOADERS, TRAIN_DATALOADERS
@@ -28,7 +29,11 @@ def __init__(self, trainer: "pl.Trainer", multiple_trainloader_mode: str = "max_
         self.multiple_trainloader_mode = multiple_trainloader_mode
 
     def on_trainer_init(
-        self, check_val_every_n_epoch: int, reload_dataloaders_every_epoch: bool, prepare_data_per_node: bool
+        self,
+        check_val_every_n_epoch: int,
+        reload_dataloaders_every_n_epochs: int,
+        reload_dataloaders_every_epoch: bool,
+        prepare_data_per_node: bool,
     ) -> None:
         self.trainer.datamodule = None
         self.trainer.prepare_data_per_node = prepare_data_per_node
@@ -39,7 +44,21 @@ def on_trainer_init(
             )
 
         self.trainer.check_val_every_n_epoch = check_val_every_n_epoch
-        self.trainer.reload_dataloaders_every_epoch = reload_dataloaders_every_epoch
+
+        if reload_dataloaders_every_epoch:
+            reload_dataloaders_every_n_epochs = int(reload_dataloaders_every_epoch)
+            rank_zero_deprecation(
+                "`reload_dataloaders_every_epoch` is deprecated in v1.4 and will be removed in v1.6."
+                " Please use `reload_dataloaders_every_n_epochs` in Trainer."
+            )
+
+        if not isinstance(reload_dataloaders_every_n_epochs, int) or (reload_dataloaders_every_n_epochs < 0):
+            raise MisconfigurationException(
+                "`reload_dataloaders_every_n_epochs` should be an int >= 0,"
+                f" got {reload_dataloaders_every_n_epochs}."
+            )
+
+        self.trainer.reload_dataloaders_every_n_epochs = reload_dataloaders_every_n_epochs
         self.trainer._is_data_prepared = False
 
     def get_profiled_train_dataloader(self, train_dataloader):

pytorch_lightning/trainer/properties.py

@@ -59,6 +59,7 @@ class TrainerProperties(ABC):
     accelerator_connector: AcceleratorConnector
     callbacks: List[Callback]
     checkpoint_connector: CheckpointConnector
+    reload_dataloaders_every_n_epochs: int
     limit_val_batches: int
     logger: LightningLoggerBase
     logger_connector: LoggerConnector
@@ -293,6 +294,12 @@ def progress_bar_dict(self) -> dict:
             )
         return {**standard_metrics, **pbar_metrics}
 
+    @property
+    def _should_reload_dl_epoch(self) -> bool:
+        """ Check if dataloader should be reloaded in the current epoch. """
+        n_epochs = self.reload_dataloaders_every_n_epochs
+        return n_epochs and (not self.current_epoch % n_epochs)
+
     @property
     def disable_validation(self) -> bool:
         """ Check if validation is disabled during training. """

pytorch_lightning/trainer/trainer.py

@@ -144,6 +144,7 @@ def __init__(
         profiler: Optional[Union[BaseProfiler, str]] = None,
         benchmark: bool = False,
         deterministic: bool = False,
+        reload_dataloaders_every_n_epochs: int = 0,
         reload_dataloaders_every_epoch: bool = False,
         auto_lr_find: Union[bool, str] = False,
         replace_sampler_ddp: bool = True,
@@ -272,8 +273,15 @@ def __init__(
             num_sanity_val_steps: Sanity check runs n validation batches before starting the training routine.
                 Set it to `-1` to run all batches in all validation dataloaders.
 
+            reload_dataloaders_every_n_epochs: Set to a non-negative integer to reload dataloaders every n epochs.
+                Default: 0
+
             reload_dataloaders_every_epoch: Set to True to reload dataloaders every epoch.
 
+                .. deprecated:: v1.4
+                    ``reload_dataloaders_every_epoch`` has been deprecated in v1.4 and will be removed in v1.6.
+                    Please use ``reload_dataloaders_every_n_epochs``.
+
             replace_sampler_ddp: Explicitly enables or disables sampler replacement. If not specified this
                 will toggled automatically when DDP is used. By default it will add ``shuffle=True`` for
                 train sampler and ``shuffle=False`` for val/test sampler. If you want to customize it,
@@ -382,7 +390,8 @@ def __init__(
 
         # init data flags
         self.data_connector.on_trainer_init(
-            check_val_every_n_epoch, reload_dataloaders_every_epoch, prepare_data_per_node
+            check_val_every_n_epoch, reload_dataloaders_every_n_epochs, reload_dataloaders_every_epoch,
+            prepare_data_per_node
         )
 
         # init training tricks

tests/core/test_datamodules.py

@@ -455,9 +455,11 @@ def transfer_batch_to_device(self, batch, device, dataloader_idx):
     assert torch.allclose(batch_gpu.targets.cpu(), torch.ones(5, 1, dtype=torch.long) * 2)
 
 
-def test_dm_reload_dataloaders_every_epoch(tmpdir):
-    """Test datamodule, where trainer argument
-    reload_dataloaders_every_epoch is set to True/False"""
+def test_dm_reload_dataloaders_every_n_epochs(tmpdir):
+    """
+    Test datamodule, where trainer argument
+    reload_dataloaders_every_n_epochs is set to a non negative integer
+    """
 
     class CustomBoringDataModule(BoringDataModule):
 
@@ -482,9 +484,9 @@ def train_dataloader(self):
 
     trainer = Trainer(
         default_root_dir=tmpdir,
-        max_epochs=2,
-        limit_train_batches=0.01,
-        reload_dataloaders_every_epoch=True,
+        max_epochs=3,
+        limit_train_batches=2,
+        reload_dataloaders_every_n_epochs=2,
     )
     trainer.fit(model, dm)
 

tests/deprecated_api/test_remove_1-6.py

@@ -88,6 +88,28 @@ def test_v1_6_0_ddp_spawn_sync_batchnorm():
         DDPSpawnPlugin(sync_batchnorm=False)
 
 
+def test_v1_6_0_reload_dataloaders_every_epoch(tmpdir):
+
+    model = BoringModel()
+
+    with pytest.deprecated_call(match='`reload_dataloaders_every_epoch` is deprecated in v1.4 and will be removed'):
+        trainer = Trainer(
+            default_root_dir=tmpdir,
+            limit_train_batches=0.3,
+            limit_val_batches=0.3,
+            reload_dataloaders_every_epoch=True,
+            max_epochs=3,
+        )
+    trainer.fit(model)
+    trainer.test()
+
+    # verify the sequence
+    calls = trainer.dev_debugger.dataloader_sequence_calls
+    expected_sequence = ['val_dataloader'] + ['train_dataloader', 'val_dataloader'] * 3 + ['test_dataloader']
+    for call, expected in zip(calls, expected_sequence):
+        assert call['name'] == expected
+
+
 def test_v1_6_0_tbptt_reduce_fx(tmpdir):
 
     class TestModel(BoringModel):

tests/trainer/test_dataloaders.py

@@ -1304,7 +1304,7 @@ def test_dataloaders_load_only_once_val_interval(tmpdir):
         limit_train_batches=10,
         limit_val_batches=10,
         val_check_interval=0.3,
-        reload_dataloaders_every_epoch=True,
+        reload_dataloaders_every_n_epochs=True,
         max_epochs=3,
     )
     trainer.fit(model)
@@ -1368,44 +1368,46 @@ def test_dataloaders_load_only_once_no_sanity_check(tmpdir):
         assert call['name'] == expected
 
 
-@mock.patch.dict(os.environ, {"PL_DEV_DEBUG": "1"})
-def test_dataloaders_load_every_epoch(tmpdir):
+@pytest.mark.parametrize("n", [1, 2])
+def test_dataloaders_load_every_n_epochs(tmpdir, n):
 
-    model = EvalModelTemplate()
+    model = BoringModel()
 
-    # logger file to get meta
     trainer = Trainer(
         default_root_dir=tmpdir,
         limit_train_batches=0.3,
         limit_val_batches=0.3,
-        reload_dataloaders_every_epoch=True,
+        reload_dataloaders_every_n_epochs=n,
         max_epochs=3,
     )
     trainer.fit(model)
     assert trainer.state.finished, f"Training failed with {trainer.state}"
 
     trainer.test()
 
-    assert len(trainer.dev_debugger.val_dataloader_calls) == 4
-    assert len(trainer.dev_debugger.train_dataloader_calls) == 3
-    assert len(trainer.dev_debugger.test_dataloader_calls) == 1
-
     # verify the sequence
     calls = trainer.dev_debugger.dataloader_sequence_calls
-    expected_sequence = [
-        'val_dataloader',
-        'train_dataloader',
-        'val_dataloader',
-        'train_dataloader',
-        'val_dataloader',
-        'train_dataloader',
-        'val_dataloader',
-        'test_dataloader',
-    ]
+    expected_sequence = ['val_dataloader']
+    if n == 1:
+        expected_sequence += ['train_dataloader', 'val_dataloader'] * 3
+    elif n == 2:
+        expected_sequence += ['train_dataloader', 'val_dataloader'] * 2
+    expected_sequence += ['test_dataloader']
+
     for call, expected in zip(calls, expected_sequence):
         assert call['name'] == expected
 
 
+@pytest.mark.parametrize("n", ['test', -1])
+def test_dataloaders_load_every_n_epochs_exception(tmpdir, n):
+
+    with pytest.raises(MisconfigurationException, match='should be an int >'):
+        Trainer(
+            default_root_dir=tmpdir,
+            reload_dataloaders_every_n_epochs=n,
+        )
+
+
 @mock.patch.dict(os.environ, {"PL_DEV_DEBUG": "1"})
 def test_dataloaders_load_every_epoch_no_sanity_check(tmpdir):
 
@@ -1426,7 +1428,7 @@ def validation_step(self, batch, batch_idx):
         limit_train_batches=0.3,
         limit_val_batches=0.3,
         num_sanity_val_steps=0,
-        reload_dataloaders_every_epoch=True,
+        reload_dataloaders_every_n_epochs=True,
         max_epochs=3,
         callbacks=[checkpoint_callback],
     )