bugfix: batch_size parameter for DataModules remaining (Lightning-Uni…

…verse#344)

* bugfix: batch_size for DataModules remaining

* Update sklearn datamodule tests

* Fix default_transforms. Keep internal for every data module

* fix typo on binary_mnist_datamodule

thanks @akihironitta

Co-authored-by: Akihiro Nitta <[email protected]>

Co-authored-by: Akihiro Nitta <[email protected]>

docs/source/introduction_guide.rst

@@ -425,7 +425,7 @@ Here's an example for logistic regression
 
     # use any numpy or sklearn dataset
     X, y = load_iris(return_X_y=True)
-    dm = SklearnDataModule(X, y)
+    dm = SklearnDataModule(X, y, batch_size=12)
 
     # build model
     model = LogisticRegression(input_dim=4, num_classes=3)
@@ -434,7 +434,7 @@ Here's an example for logistic regression
     trainer = pl.Trainer(tpu_cores=8, precision=16)
     trainer.fit(model, dm.train_dataloader(), dm.val_dataloader())
 
-    trainer.test(test_dataloaders=dm.test_dataloader(batch_size=12))
+    trainer.test(test_dataloaders=dm.test_dataloader())
 
 Any input will be flattened across all dimensions except the first one (batch).
 This means images, sound, etc... work out of the box.

pl_bolts/datamodules/cityscapes_datamodule.py

@@ -111,8 +111,8 @@ def train_dataloader(self):
         """
         Cityscapes train set
         """
-        transforms = self.train_transforms or self.default_transforms()
-        target_transforms = self.target_transforms or self.default_target_transforms()
+        transforms = self.train_transforms or self._default_transforms()
+        target_transforms = self.target_transforms or self._default_target_transforms()
 
         dataset = Cityscapes(self.data_dir,
                              split='train',
@@ -136,8 +136,8 @@ def val_dataloader(self):
         """
         Cityscapes val set
         """
-        transforms = self.val_transforms or self.default_transforms()
-        target_transforms = self.target_transforms or self.default_target_transforms()
+        transforms = self.val_transforms or self._default_transforms()
+        target_transforms = self.target_transforms or self._default_target_transforms()
 
         dataset = Cityscapes(self.data_dir,
                              split='val',
@@ -161,8 +161,8 @@ def test_dataloader(self):
         """
         Cityscapes test set
         """
-        transforms = self.test_transforms or self.default_transforms()
-        target_transforms = self.target_transforms or self.default_target_transforms()
+        transforms = self.test_transforms or self._default_transforms()
+        target_transforms = self.target_transforms or self._default_target_transforms()
 
         dataset = Cityscapes(self.data_dir,
                              split='test',
@@ -181,7 +181,7 @@ def test_dataloader(self):
         )
         return loader
 
-    def default_transforms(self):
+    def _default_transforms(self):
         cityscapes_transforms = transform_lib.Compose([
             transform_lib.ToTensor(),
             transform_lib.Normalize(
@@ -191,7 +191,7 @@ def default_transforms(self):
         ])
         return cityscapes_transforms
 
-    def default_target_transforms(self):
+    def _default_target_transforms(self):
         cityscapes_target_trasnforms = transform_lib.Compose([
             transform_lib.ToTensor(),
             transform_lib.Lambda(lambda t: t.squeeze())

pl_bolts/datamodules/kitti_datamodule.py

@@ -74,14 +74,8 @@ def __init__(
         self.num_workers = num_workers
         self.seed = seed
 
-        self.default_transforms = transforms.Compose([
-            transforms.ToTensor(),
-            transforms.Normalize(mean=[0.35675976, 0.37380189, 0.3764753],
-                                 std=[0.32064945, 0.32098866, 0.32325324])
-        ])
-
         # split into train, val, test
-        kitti_dataset = KittiDataset(self.data_dir, transform=self.default_transforms)
+        kitti_dataset = KittiDataset(self.data_dir, transform=self._default_transforms())
 
         val_len = round(val_split * len(kitti_dataset))
         test_len = round(test_split * len(kitti_dataset))
@@ -111,3 +105,11 @@ def test_dataloader(self):
                             shuffle=False,
                             num_workers=self.num_workers)
         return loader
+
+    def _default_transforms(self):
+        kitti_transforms = transforms.Compose([
+            transforms.ToTensor(),
+            transforms.Normalize(mean=[0.35675976, 0.37380189, 0.3764753],
+                                 std=[0.32064945, 0.32098866, 0.32325324])
+        ])
+        return kitti_transforms

pl_bolts/datamodules/sklearn_datamodule.py

@@ -118,22 +118,22 @@ class SklearnDataModule(LightningDataModule):
         >>> from pl_bolts.datamodules import SklearnDataModule
         ...
         >>> X, y = load_boston(return_X_y=True)
-        >>> loaders = SklearnDataModule(X, y)
+        >>> loaders = SklearnDataModule(X, y, batch_size=32)
         ...
         >>> # train set
-        >>> train_loader = loaders.train_dataloader(batch_size=32)
+        >>> train_loader = loaders.train_dataloader()
         >>> len(train_loader.dataset)
         355
         >>> len(train_loader)
         11
         >>> # validation set
-        >>> val_loader = loaders.val_dataloader(batch_size=32)
+        >>> val_loader = loaders.val_dataloader()
         >>> len(val_loader.dataset)
         100
         >>> len(val_loader)
         3
         >>> # test set
-        >>> test_loader = loaders.test_dataloader(batch_size=32)
+        >>> test_loader = loaders.test_dataloader()
         >>> len(test_loader.dataset)
         51
         >>> len(test_loader)
@@ -150,12 +150,14 @@ def __init__(
             num_workers=2,
             random_state=1234,
             shuffle=True,
+            batch_size: int = 16,
             *args,
             **kwargs,
     ):
 
         super().__init__(*args, **kwargs)
         self.num_workers = num_workers
+        self.batch_size = batch_size
 
         # shuffle x and y
         if shuffle and _SKLEARN_AVAILABLE:
@@ -193,32 +195,32 @@ def _init_datasets(self, X, y, x_val, y_val, x_test, y_test):
         self.val_dataset = SklearnDataset(x_val, y_val)
         self.test_dataset = SklearnDataset(x_test, y_test)
 
-    def train_dataloader(self, batch_size: int = 16):
+    def train_dataloader(self):
         loader = DataLoader(
             self.train_dataset,
-            batch_size=batch_size,
+            batch_size=self.batch_size,
             shuffle=True,
             num_workers=self.num_workers,
             drop_last=True,
             pin_memory=True
         )
         return loader
 
-    def val_dataloader(self, batch_size: int = 16):
+    def val_dataloader(self):
         loader = DataLoader(
             self.val_dataset,
-            batch_size=batch_size,
+            batch_size=self.batch_size,
             shuffle=False,
             num_workers=self.num_workers,
             drop_last=True,
             pin_memory=True
         )
         return loader
 
-    def test_dataloader(self, batch_size: int = 16):
+    def test_dataloader(self):
         loader = DataLoader(
             self.test_dataset,
-            batch_size=batch_size,
+            batch_size=self.batch_size,
             shuffle=False,
             num_workers=self.num_workers,
             drop_last=True,

pl_bolts/datamodules/ssl_imagenet_datamodule.py

@@ -26,6 +26,7 @@ def __init__(
             data_dir,
             meta_dir=None,
             num_workers=16,
+            batch_size: int = 32,
             *args,
             **kwargs,
     ):
@@ -39,6 +40,7 @@ def __init__(
         self.data_dir = data_dir
         self.num_workers = num_workers
         self.meta_dir = meta_dir
+        self.batch_size = batch_size
 
     @property
     def num_classes(self):
@@ -74,7 +76,7 @@ def prepare_data(self):
                 UnlabeledImagenet.generate_meta_bins(path)
                 """)
 
-    def train_dataloader(self, batch_size, num_images_per_class=-1, add_normalize=False):
+    def train_dataloader(self, num_images_per_class=-1, add_normalize=False):
         transforms = self._default_transforms() if self.train_transforms is None else self.train_transforms
 
         dataset = UnlabeledImagenet(self.data_dir,
@@ -84,15 +86,15 @@ def train_dataloader(self, batch_size, num_images_per_class=-1, add_normalize=Fa
                                     transform=transforms)
         loader = DataLoader(
             dataset,
-            batch_size=batch_size,
+            batch_size=self.batch_size,
             shuffle=True,
             num_workers=self.num_workers,
             drop_last=True,
             pin_memory=True
         )
         return loader
 
-    def val_dataloader(self, batch_size, num_images_per_class=50, add_normalize=False):
+    def val_dataloader(self, num_images_per_class=50, add_normalize=False):
         transforms = self._default_transforms() if self.val_transforms is None else self.val_transforms
 
         dataset = UnlabeledImagenet(self.data_dir,
@@ -102,14 +104,14 @@ def val_dataloader(self, batch_size, num_images_per_class=50, add_normalize=Fals
                                     transform=transforms)
         loader = DataLoader(
             dataset,
-            batch_size=batch_size,
+            batch_size=self.batch_size,
             shuffle=False,
             num_workers=self.num_workers,
             pin_memory=True
         )
         return loader
 
-    def test_dataloader(self, batch_size, num_images_per_class, add_normalize=False):
+    def test_dataloader(self, num_images_per_class, add_normalize=False):
         transforms = self._default_transforms() if self.test_transforms is None else self.test_transforms
 
         dataset = UnlabeledImagenet(self.data_dir,
@@ -119,7 +121,7 @@ def test_dataloader(self, batch_size, num_images_per_class, add_normalize=False)
                                     transform=transforms)
         loader = DataLoader(
             dataset,
-            batch_size=batch_size,
+            batch_size=self.batch_size,
             shuffle=False,
             num_workers=self.num_workers,
             drop_last=True,

pl_bolts/datamodules/stl10_datamodule.py

@@ -105,7 +105,7 @@ def train_dataloader(self):
         """
         Loads the 'unlabeled' split minus a portion set aside for validation via `unlabeled_val_split`.
         """
-        transforms = self.default_transforms() if self.train_transforms is None else self.train_transforms
+        transforms = self._default_transforms() if self.train_transforms is None else self.train_transforms
 
         dataset = STL10(self.data_dir, split='unlabeled', download=False, transform=transforms)
         train_length = len(dataset)
@@ -132,7 +132,7 @@ def train_dataloader_mixed(self):
             batch_size: the batch size
             transforms: a sequence of transforms
         """
-        transforms = self.default_transforms() if self.train_transforms is None else self.train_transforms
+        transforms = self._default_transforms() if self.train_transforms is None else self.train_transforms
 
         unlabeled_dataset = STL10(self.data_dir,
                                   split='unlabeled',
@@ -170,7 +170,7 @@ def val_dataloader(self):
             batch_size: the batch size
             transforms: a sequence of transforms
         """
-        transforms = self.default_transforms() if self.val_transforms is None else self.val_transforms
+        transforms = self._default_transforms() if self.val_transforms is None else self.val_transforms
 
         dataset = STL10(self.data_dir, split='unlabeled', download=False, transform=transforms)
         train_length = len(dataset)
@@ -202,7 +202,7 @@ def val_dataloader_mixed(self):
             batch_size: the batch size
             transforms: a sequence of transforms
         """
-        transforms = self.default_transforms() if self.val_transforms is None else self.val_transforms
+        transforms = self._default_transforms() if self.val_transforms is None else self.val_transforms
         unlabeled_dataset = STL10(self.data_dir,
                                   split='unlabeled',
                                   download=False,
@@ -237,7 +237,7 @@ def test_dataloader(self):
             batch_size: the batch size
             transforms: the transforms
         """
-        transforms = self.default_transforms() if self.test_transforms is None else self.test_transforms
+        transforms = self._default_transforms() if self.test_transforms is None else self.test_transforms
 
         dataset = STL10(self.data_dir, split='test', download=False, transform=transforms)
         loader = DataLoader(
@@ -251,7 +251,7 @@ def test_dataloader(self):
         return loader
 
     def train_dataloader_labeled(self):
-        transforms = self.default_transforms() if self.val_transforms is None else self.val_transforms
+        transforms = self._default_transforms() if self.val_transforms is None else self.val_transforms
 
         dataset = STL10(self.data_dir, split='train', download=False, transform=transforms)
         train_length = len(dataset)
@@ -268,7 +268,7 @@ def train_dataloader_labeled(self):
         return loader
 
     def val_dataloader_labeled(self):
-        transforms = self.default_transforms() if self.val_transforms is None else self.val_transforms
+        transforms = self._default_transforms() if self.val_transforms is None else self.val_transforms
         dataset = STL10(self.data_dir,
                         split='train',
                         download=False,
@@ -288,7 +288,7 @@ def val_dataloader_labeled(self):
         )
         return loader
 
-    def default_transforms(self):
+    def _default_transforms(self):
         data_transforms = transform_lib.Compose([
             transform_lib.ToTensor(),
             stl10_normalization()

pl_bolts/models/regression/linear_regression.py

@@ -131,9 +131,6 @@ def cli_main():
             'You want to use `sklearn` which is not installed yet, install it with `pip install sklearn`.'
         ) from err
 
-    X, y = load_boston(return_X_y=True)  # these are numpy arrays
-    loaders = SklearnDataModule(X, y)
-
     # args
     parser = ArgumentParser()
     parser = LinearRegression.add_model_specific_args(parser)
@@ -144,9 +141,13 @@ def cli_main():
     model = LinearRegression(input_dim=13, l1_strength=1, l2_strength=1)
     # model = LinearRegression(**vars(args))
 
+    # data
+    X, y = load_boston(return_X_y=True)  # these are numpy arrays
+    loaders = SklearnDataModule(X, y, batch_size=args.batch_size)
+
     # train
     trainer = pl.Trainer.from_argparse_args(args)
-    trainer.fit(model, loaders.train_dataloader(args.batch_size), loaders.val_dataloader(args.batch_size))
+    trainer.fit(model, loaders.train_dataloader(), loaders.val_dataloader())
 
 
 if __name__ == '__main__':

pl_bolts/models/regression/logistic_regression.py

@@ -137,9 +137,6 @@ def cli_main():
             'You want to use `sklearn` which is not installed yet, install it with `pip install sklearn`.'
         ) from err
 
-    X, y = load_iris(return_X_y=True)
-    loaders = SklearnDataModule(X, y)
-
     # args
     parser = ArgumentParser()
     parser = LogisticRegression.add_model_specific_args(parser)
@@ -150,9 +147,13 @@ def cli_main():
     # model = LogisticRegression(**vars(args))
     model = LogisticRegression(input_dim=4, num_classes=3, l1_strength=0.01, learning_rate=0.01)
 
+    # data
+    X, y = load_iris(return_X_y=True)
+    loaders = SklearnDataModule(X, y, batch_size=args.batch_size)
+
     # train
     trainer = pl.Trainer.from_argparse_args(args)
-    trainer.fit(model, loaders.train_dataloader(args.batch_size), loaders.val_dataloader(args.batch_size))
+    trainer.fit(model, loaders.train_dataloader(), loaders.val_dataloader())
 
 
 if __name__ == '__main__':