flake8 ++

CHANGELOG.md

@@ -40,12 +40,6 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Support number for logging with `sync_dist=True` ([#5080](https://github.com/PyTorchLightning/pytorch-lightning/pull/5080)
 - Added offset logging step when resuming for Wandb logger ([#5050](https://github.com/PyTorchLightning/pytorch-lightning/pull/5050)
 
-### Changed
-
-
-### Deprecated
-
-
 ### Removed
 
 - `enable_pl_optimizer=False` by default to temporarily fix AMP issues ([#5163](https://github.com/PyTorchLightning/pytorch-lightning/pull/5163)

pl_examples/__init__.py

@@ -2,12 +2,12 @@
 
 from pytorch_lightning.utilities import _module_available
 
-EXAMPLES_ROOT = os.path.dirname(__file__)
-PACKAGE_ROOT = os.path.dirname(EXAMPLES_ROOT)
-DATASETS_PATH = os.path.join(PACKAGE_ROOT, 'Datasets')
+_EXAMPLES_ROOT = os.path.dirname(__file__)
+_PACKAGE_ROOT = os.path.dirname(_EXAMPLES_ROOT)
+_DATASETS_PATH = os.path.join(_PACKAGE_ROOT, 'Datasets')
 
-TORCHVISION_AVAILABLE = _module_available("torchvision")
-DALI_AVAILABLE = _module_available("nvidia.dali")
+_TORCHVISION_AVAILABLE = _module_available("torchvision")
+_DALI_AVAILABLE = _module_available("nvidia.dali")
 
 
 LIGHTNING_LOGO = """

pl_examples/basic_examples/autoencoder.py

@@ -21,9 +21,9 @@
 from torch.utils.data import random_split
 
 import pytorch_lightning as pl
-from pl_examples import TORCHVISION_AVAILABLE, cli_lightning_logo
+from pl_examples import _TORCHVISION_AVAILABLE, cli_lightning_logo
 
-if TORCHVISION_AVAILABLE:
+if _TORCHVISION_AVAILABLE:
     from torchvision.datasets.mnist import MNIST
     from torchvision import transforms
 else:

pl_examples/basic_examples/backbone_image_classifier.py

@@ -19,9 +19,9 @@
 from torch.utils.data import DataLoader, random_split
 
 import pytorch_lightning as pl
-from pl_examples import DATASETS_PATH, TORCHVISION_AVAILABLE, cli_lightning_logo
+from pl_examples import _DATASETS_PATH, _TORCHVISION_AVAILABLE, cli_lightning_logo
 
-if TORCHVISION_AVAILABLE:
+if _TORCHVISION_AVAILABLE:
     from torchvision.datasets.mnist import MNIST
     from torchvision import transforms
 else:
@@ -111,8 +111,8 @@ def cli_main():
     # ------------
     # data
     # ------------
-    dataset = MNIST(DATASETS_PATH, train=True, download=True, transform=transforms.ToTensor())
-    mnist_test = MNIST(DATASETS_PATH, train=False, download=True, transform=transforms.ToTensor())
+    dataset = MNIST(_DATASETS_PATH, train=True, download=True, transform=transforms.ToTensor())
+    mnist_test = MNIST(_DATASETS_PATH, train=False, download=True, transform=transforms.ToTensor())
     mnist_train, mnist_val = random_split(dataset, [55000, 5000])
 
     train_loader = DataLoader(mnist_train, batch_size=args.batch_size)

pl_examples/basic_examples/dali_image_classifier.py

@@ -23,15 +23,15 @@
 from torch.utils.data import random_split
 
 import pytorch_lightning as pl
-from pl_examples import TORCHVISION_AVAILABLE, DALI_AVAILABLE, cli_lightning_logo
+from pl_examples import _TORCHVISION_AVAILABLE, _DALI_AVAILABLE, cli_lightning_logo
 
-if TORCHVISION_AVAILABLE:
+if _TORCHVISION_AVAILABLE:
     from torchvision.datasets.mnist import MNIST
     from torchvision import transforms
 else:
     from tests.base.datasets import MNIST
 
-if DALI_AVAILABLE:
+if _DALI_AVAILABLE:
     from nvidia.dali import ops
     from nvidia.dali.pipeline import Pipeline
     from nvidia.dali.plugin.pytorch import DALIClassificationIterator
@@ -166,7 +166,7 @@ def add_model_specific_args(parent_parser):
 
 
 def cli_main():
-    if not DALI_AVAILABLE:
+    if not _DALI_AVAILABLE:
         return
 
     pl.seed_everything(1234)

pl_examples/basic_examples/mnist_datamodule.py

@@ -16,10 +16,10 @@
 
 from torch.utils.data import DataLoader, random_split
 
-from pl_examples import DATASETS_PATH, TORCHVISION_AVAILABLE
+from pl_examples import _DATASETS_PATH, _TORCHVISION_AVAILABLE
 from pytorch_lightning import LightningDataModule
 
-if TORCHVISION_AVAILABLE:
+if _TORCHVISION_AVAILABLE:
     from torchvision import transforms as transform_lib
     from torchvision.datasets import MNIST
 else:
@@ -38,7 +38,7 @@ class MNISTDataModule(LightningDataModule):
 
     def __init__(
         self,
-        data_dir: str = DATASETS_PATH,
+        data_dir: str = _DATASETS_PATH,
         val_split: int = 5000,
         num_workers: int = 16,
         normalize: bool = False,
@@ -123,7 +123,7 @@ def test_dataloader(self):
 
     @property
     def default_transforms(self):
-        if not TORCHVISION_AVAILABLE:
+        if not _TORCHVISION_AVAILABLE:
             return None
         if self.normalize:
             mnist_transforms = transform_lib.Compose(

pl_examples/domain_templates/semantic_segmentation.py

@@ -67,8 +67,8 @@ class KITTI(Dataset):
     encoded using `encode_segmap`, and given `transform` (if any) are applied to the image only
     (mask does not usually require transforms, but they can be implemented in a similar way).
 
-    >>> from pl_examples import DATASETS_PATH
-    >>> dataset_path = os.path.join(DATASETS_PATH, "Kitti")
+    >>> from pl_examples import _DATASETS_PATH
+    >>> dataset_path = os.path.join(_DATASETS_PATH, "Kitti")
     >>> _create_synth_kitti_dataset(dataset_path, image_dims=(1024, 512))
     >>> KITTI(dataset_path, 'train')  # doctest: +ELLIPSIS +NORMALIZE_WHITESPACE
     <...semantic_segmentation.KITTI object at ...>
@@ -161,8 +161,8 @@ class SegModel(pl.LightningModule):
 
     Adam optimizer is used along with Cosine Annealing learning rate scheduler.
 
-    >>> from pl_examples import DATASETS_PATH
-    >>> dataset_path = os.path.join(DATASETS_PATH, "Kitti")
+    >>> from pl_examples import _DATASETS_PATH
+    >>> dataset_path = os.path.join(_DATASETS_PATH, "Kitti")
     >>> _create_synth_kitti_dataset(dataset_path, image_dims=(1024, 512))
     >>> SegModel(dataset_path)  # doctest: +ELLIPSIS +NORMALIZE_WHITESPACE
     SegModel(

pl_examples/test_examples.py

@@ -19,7 +19,7 @@
 import pytest
 import torch
 
-from pl_examples import DALI_AVAILABLE
+from pl_examples import _DALI_AVAILABLE
 
 ARGS_DEFAULT = """
 --default_root_dir %(tmpdir)s \
@@ -104,7 +104,7 @@ def test_examples_cpu(tmpdir, import_cli, cli_args):
         module.cli_main()
 
 
-@pytest.mark.skipif(not DALI_AVAILABLE, reason="Nvidia DALI required")
+@pytest.mark.skipif(not _DALI_AVAILABLE, reason="Nvidia DALI required")
 @pytest.mark.skipif(not torch.cuda.is_available(), reason="test requires GPU machine")
 @pytest.mark.skipif(platform.system() != 'Linux', reason='Only applies to Linux platform.')
 @pytest.mark.parametrize('cli_args', [ARGS_GPU])

pytorch_lightning/__init__.py

@@ -40,8 +40,8 @@
 _logger.addHandler(python_logging.StreamHandler())
 _logger.setLevel(python_logging.INFO)
 
-PACKAGE_ROOT = os.path.dirname(__file__)
-PROJECT_ROOT = os.path.dirname(PACKAGE_ROOT)
+_PACKAGE_ROOT = os.path.dirname(__file__)
+_PROJECT_ROOT = os.path.dirname(_PACKAGE_ROOT)
 
 try:
     # This variable is injected in the __builtins__ by the build

pytorch_lightning/loggers/wandb.py

@@ -162,7 +162,9 @@ def log_metrics(self, metrics: Dict[str, float], step: Optional[int] = None) ->
 
         metrics = self._add_prefix(metrics)
         if step is not None and step + self._step_offset < self.experiment.step:
-            self.warning_cache.warn('Trying to log at a previous step. Use `commit=False` when logging metrics manually.')
+            self.warning_cache.warn(
+                'Trying to log at a previous step. Use `commit=False` when logging metrics manually.'
+            )
         self.experiment.log(metrics, step=(step + self._step_offset) if step is not None else None)
 
     @property

pytorch_lightning/setup_tools.py

@@ -19,7 +19,7 @@
 from urllib.request import Request, urlopen
 import warnings
 
-from pytorch_lightning import __homepage__, __version__, PROJECT_ROOT
+from pytorch_lightning import __homepage__, __version__, _PROJECT_ROOT
 
 _PATH_BADGES = os.path.join('.', 'docs', 'source', '_images', 'badges')
 # badge to download
@@ -37,7 +37,7 @@
 def _load_requirements(path_dir: str , file_name: str = 'requirements.txt', comment_char: str = '#') -> List[str]:
     """Load requirements from a file
 
-    >>> _load_requirements(PROJECT_ROOT)  # doctest: +ELLIPSIS +NORMALIZE_WHITESPACE
+    >>> _load_requirements(_PROJECT_ROOT)  # doctest: +ELLIPSIS +NORMALIZE_WHITESPACE
     ['numpy...', 'torch...', ...]
     """
     with open(os.path.join(path_dir, file_name), 'r') as file:
@@ -155,7 +155,7 @@ def _download_badge(url_badge: str, badge_name: str, target_dir: str) -> str:
 def _load_long_description(path_dir: str) -> str:
     """Load readme as decribtion
 
-    >>> _load_long_description(PROJECT_ROOT)  # doctest: +ELLIPSIS +NORMALIZE_WHITESPACE
+    >>> _load_long_description(_PROJECT_ROOT)  # doctest: +ELLIPSIS +NORMALIZE_WHITESPACE
     '<div align="center">...'
     """
     path_readme = os.path.join(path_dir, "README.md")

pytorch_lightning/utilities/__init__.py

@@ -21,7 +21,7 @@
 import numpy
 import torch
 
-from pytorch_lightning.utilities.apply_func import move_data_to_device  # noqa: F401
+from pytorch_lightning.utilities.apply_func import _TORCHTEXT_AVAILABLE, move_data_to_device  # noqa: F401
 from pytorch_lightning.utilities.distributed import (  # noqa: F401
     AllGatherGrad,
     rank_zero_info,
@@ -59,14 +59,14 @@ def _module_available(module_path: str) -> bool:
 _OMEGACONF_AVAILABLE = _module_available("omegaconf")
 _HYDRA_AVAILABLE = _module_available("hydra")
 _HOROVOD_AVAILABLE = _module_available("horovod.torch")
-
-_TPU_AVAILABLE = XLADeviceUtils.tpu_device_exists()
 _FAIRSCALE_AVAILABLE = platform.system() != 'Windows' and _module_available('fairscale.nn.data_parallel')
 _RPC_AVAILABLE = platform.system() != 'Windows' and _module_available('torch.distributed.rpc')
 _GROUP_AVAILABLE = platform.system() != 'Windows' and _module_available('torch.distributed.group')
 _FAIRSCALE_PIPE_AVAILABLE = _FAIRSCALE_AVAILABLE and LooseVersion(torch.__version__) >= LooseVersion("1.6.0")
 _BOLTS_AVAILABLE = _module_available('pl_bolts')
 
+_TPU_AVAILABLE = XLADeviceUtils.tpu_device_exists()
+
 FLOAT16_EPSILON = numpy.finfo(numpy.float16).eps
 FLOAT32_EPSILON = numpy.finfo(numpy.float32).eps
 FLOAT64_EPSILON = numpy.finfo(numpy.float64).eps

pytorch_lightning/utilities/apply_func.py

@@ -11,7 +11,6 @@
 # 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.
-
 import importlib
 from abc import ABC
 from collections.abc import Mapping, Sequence
@@ -20,8 +19,9 @@
 
 import torch
 
-TORCHTEXT_AVAILABLE = importlib.util.find_spec("torchtext") is not None
-if TORCHTEXT_AVAILABLE:
+_TORCHTEXT_AVAILABLE = importlib.util.find_spec("torchtext") is not None
+
+if _TORCHTEXT_AVAILABLE:
     from torchtext.data import Batch
 else:
     Batch = type(None)
@@ -109,7 +109,7 @@ def move_data_to_device(batch: Any, device: torch.device):
 
     def batch_to(data):
         # try to move torchtext data first
-        if TORCHTEXT_AVAILABLE and isinstance(data, Batch):
+        if _TORCHTEXT_AVAILABLE and isinstance(data, Batch):
 
             # Shallow copy because each Batch has a reference to Dataset which contains all examples
             device_data = copy(data)
@@ -123,5 +123,5 @@ def batch_to(data):
         kwargs = dict(non_blocking=True) if isinstance(data, torch.Tensor) else {}
         return data.to(device, **kwargs)
 
-    dtype = (TransferableDataType, Batch) if TORCHTEXT_AVAILABLE else TransferableDataType
+    dtype = (TransferableDataType, Batch) if _TORCHTEXT_AVAILABLE else TransferableDataType
     return apply_to_collection(batch, dtype=dtype, function=batch_to)

tests/__init__.py

@@ -15,17 +15,17 @@
 
 import numpy as np
 
-TEST_ROOT = os.path.dirname(__file__)
-PROJECT_ROOT = os.path.dirname(TEST_ROOT)
-TEMP_PATH = os.path.join(PROJECT_ROOT, 'test_temp')
+_TEST_ROOT = os.path.dirname(__file__)
+_PROJECT_ROOT = os.path.dirname(_TEST_ROOT)
+_TEMP_PATH = os.path.join(_PROJECT_ROOT, 'test_temp')
 
 # todo: this setting `PYTHONPATH` may not be used by other evns like Conda for import packages
-if PROJECT_ROOT not in os.getenv('PYTHONPATH', ""):
+if _PROJECT_ROOT not in os.getenv('PYTHONPATH', ""):
     splitter = ":" if os.environ.get("PYTHONPATH", "") else ""
-    os.environ['PYTHONPATH'] = f'{PROJECT_ROOT}{splitter}{os.environ.get("PYTHONPATH", "")}'
+    os.environ['PYTHONPATH'] = f'{_PROJECT_ROOT}{splitter}{os.environ.get("PYTHONPATH", "")}'
 
 # generate a list of random seeds for each test
 RANDOM_PORTS = list(np.random.randint(12000, 19000, 1000))
 
-if not os.path.isdir(TEMP_PATH):
-    os.mkdir(TEMP_PATH)
+if not os.path.isdir(_TEMP_PATH):
+    os.mkdir(_TEMP_PATH)

tests/base/datasets.py

@@ -22,10 +22,10 @@
 from torch import Tensor
 from torch.utils.data import Dataset
 
-from tests import PROJECT_ROOT
+from tests import _PROJECT_ROOT
 
 #: local path to test datasets
-PATH_DATASETS = os.path.join(PROJECT_ROOT, 'Datasets')
+PATH_DATASETS = os.path.join(_PROJECT_ROOT, 'Datasets')
 
 
 class MNIST(Dataset):

tests/base/develop_utils.py

@@ -19,7 +19,7 @@
 from pytorch_lightning import seed_everything
 from pytorch_lightning.callbacks import ModelCheckpoint
 from pytorch_lightning.loggers import TensorBoardLogger, TestTubeLogger
-from tests import TEMP_PATH, RANDOM_PORTS
+from tests import _TEMP_PATH, RANDOM_PORTS
 from tests.base.model_template import EvalModelTemplate
 
 
@@ -63,7 +63,7 @@ def get_data_path(expt_logger, path_dir=None):
         if hasattr(expt_logger, 'save_dir') and expt_logger.save_dir:
             path_dir = expt_logger.save_dir
         else:
-            path_dir = TEMP_PATH
+            path_dir = _TEMP_PATH
     path_expt = os.path.join(path_dir, name, 'version_%s' % version)
 
     # try if the new sub-folder exists, typical case for test-tube

tests/deprecated_api/test_remove_1-3.py

@@ -21,7 +21,6 @@
 from pytorch_lightning import LightningModule, Trainer
 from pytorch_lightning.callbacks import EarlyStopping, ModelCheckpoint
 from pytorch_lightning.profiler.profilers import PassThroughProfiler, SimpleProfiler
-from tests.deprecated_api import _soft_unimport_module
 
 
 def test_v1_3_0_deprecated_arguments(tmpdir):

tests/loggers/test_all.py

@@ -368,7 +368,7 @@ def test_logger_with_prefix_all(tmpdir, monkeypatch):
         logger.experiment.log.assert_called_once_with({"tmp-test": 1.0}, global_step=0)
 
     # WandB
-    with mock.patch('pytorch_lightning.loggers.wandb.wandb'):
+    with mock.patch('pytorch_lightning.loggers.wandb.wandb') as wandb:
         logger = _instantiate_logger(WandbLogger, save_idr=tmpdir, prefix=prefix)
         wandb.run = None
         wandb.init().step = 0

tests/models/test_horovod.py

@@ -44,9 +44,9 @@
     from horovod.common.util import nccl_built
     nccl_built()
 except (ImportError, ModuleNotFoundError, AttributeError):
-    HOROVOD_NCCL_AVAILABLE = False
+    _HOROVOD_NCCL_AVAILABLE = False
 finally:
-    HOROVOD_NCCL_AVAILABLE = True
+    _HOROVOD_NCCL_AVAILABLE = True
 
 
 def _run_horovod(trainer_options, on_gpu=False):
@@ -105,7 +105,7 @@ def test_horovod_cpu_implicit(enable_pl_optimizer, tmpdir):
 
 
 @pytest.mark.skipif(platform.system() == "Windows", reason="Horovod is not supported on Windows")
-@pytest.mark.skipif(not HOROVOD_NCCL_AVAILABLE, reason="test requires Horovod with NCCL support")
+@pytest.mark.skipif(not _HOROVOD_NCCL_AVAILABLE, reason="test requires Horovod with NCCL support")
 @pytest.mark.skipif(torch.cuda.device_count() < 2, reason="test requires multi-GPU machine")
 def test_horovod_multi_gpu(tmpdir):
     """Test Horovod with multi-GPU support."""
@@ -125,7 +125,7 @@ def test_horovod_multi_gpu(tmpdir):
 
 
 @pytest.mark.skipif(platform.system() == "Windows", reason="Horovod is not supported on Windows")
-@pytest.mark.skipif(not HOROVOD_NCCL_AVAILABLE, reason="test requires Horovod with NCCL support")
+@pytest.mark.skipif(not _HOROVOD_NCCL_AVAILABLE, reason="test requires Horovod with NCCL support")
 @pytest.mark.skipif(torch.cuda.device_count() < 2, reason="test requires multi-GPU machine")
 @pytest.mark.skipif(not _APEX_AVAILABLE, reason="test requires apex")
 def test_horovod_apex(tmpdir):
@@ -149,7 +149,7 @@ def test_horovod_apex(tmpdir):
 
 @pytest.mark.skip(reason="Skip till Horovod fixes integration with Native torch.cuda.amp")
 @pytest.mark.skipif(platform.system() == "Windows", reason="Horovod is not supported on Windows")
-@pytest.mark.skipif(not HOROVOD_NCCL_AVAILABLE, reason="test requires Horovod with NCCL support")
+@pytest.mark.skipif(not _HOROVOD_NCCL_AVAILABLE, reason="test requires Horovod with NCCL support")
 @pytest.mark.skipif(torch.cuda.device_count() < 2, reason="test requires multi-GPU machine")
 @pytest.mark.skipif(not _NATIVE_AMP_AVAILABLE, reason="test requires torch.cuda.amp")
 def test_horovod_amp(tmpdir):
@@ -172,7 +172,7 @@ def test_horovod_amp(tmpdir):
 
 
 @pytest.mark.skipif(platform.system() == "Windows", reason="Horovod is not supported on Windows")
-@pytest.mark.skipif(not HOROVOD_NCCL_AVAILABLE, reason="test requires Horovod with NCCL support")
+@pytest.mark.skipif(not _HOROVOD_NCCL_AVAILABLE, reason="test requires Horovod with NCCL support")
 @pytest.mark.skipif(not torch.cuda.is_available(), reason="test requires GPU machine")
 def test_horovod_transfer_batch_to_gpu(tmpdir):
     class TestTrainingStepModel(EvalModelTemplate):