CI: add flake8 (#4239)

.github/workflows/code-formatting.yml

@@ -25,7 +25,7 @@ jobs:
 
   python-types:
     name: Python static type checking with Pyright
-    runs-on: ubuntu-18.04
+    runs-on: ubuntu-20.04
 
     # Timeout: https://stackoverflow.com/a/59076067/4521646
     timeout-minutes: 15
@@ -71,3 +71,24 @@ jobs:
       - name: Run type checking
         run: |
           $(npm bin)/pyright --project .pyrightconfig.json
+
+  python-pep8:
+    name: Python formatting PEP8
+    runs-on: ubuntu-20.04
+
+    # Timeout: https://stackoverflow.com/a/59076067/4521646
+    timeout-minutes: 10
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+      - uses: actions/setup-python@v2
+        with:
+          python-version: 3.7
+
+      - name: Install dependencies
+        run: |
+          pip install flake8
+
+      - name: Run checking
+        run: |
+          flake8 .

pytorch_lightning/accelerators/ddp2_accelerator.py

@@ -16,7 +16,9 @@
 
 import torch
 import torch.distributed as torch_distrib
+
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
+from pytorch_lightning.core.lightning import LightningModule
 from pytorch_lightning.core.step_result import Result
 from pytorch_lightning.distributed.dist import LightningDistributed
 from pytorch_lightning import _logger as log
@@ -191,14 +193,14 @@ def ddp_train(self, process_idx, mp_queue, model):
         return results
 
     def configure_ddp(
-        self, model: "LightningModule", device_ids: List[int]
+        self, model: LightningModule, device_ids: List[int]
     ) -> DistributedDataParallel:
         model = LightningDistributedDataParallel(
             model, device_ids=device_ids, find_unused_parameters=True
         )
         return model
 
-    def configure_sync_batchnorm(self, model: "LightningModule") -> "LightningModule":
+    def configure_sync_batchnorm(self, model: LightningModule) -> LightningModule:
         """
         Add global batchnorm for a model spread across multiple GPUs and nodes.
 

pytorch_lightning/accelerators/ddp_accelerator.py

@@ -18,21 +18,21 @@
 import sys
 from os.path import abspath
 from time import sleep
-from typing import Optional
-import numpy as np
+from typing import Optional, List
 
+import numpy as np
 
 from pytorch_lightning import _logger as log
-from pytorch_lightning.utilities.distributed import find_free_network_port
 from pytorch_lightning.accelerators.accelerator import Accelerator
-from pytorch_lightning.utilities.distributed import rank_zero_only
-from pytorch_lightning.utilities import AMPType
-from pytorch_lightning.utilities.seed import seed_everything
+from pytorch_lightning.core.lightning import LightningModule
 from pytorch_lightning.distributed.dist import LightningDistributed
-from pytorch_lightning.utilities.exceptions import MisconfigurationException
 from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
+from pytorch_lightning.utilities import AMPType
+from pytorch_lightning.utilities.distributed import find_free_network_port
+from pytorch_lightning.utilities.distributed import rank_zero_only
+from pytorch_lightning.utilities.exceptions import MisconfigurationException
+from pytorch_lightning.utilities.seed import seed_everything
 from torch.nn.parallel import DistributedDataParallel
-from typing import List
 
 
 try:
@@ -284,14 +284,14 @@ def ddp_train(self, process_idx, model):
         return results
 
     def configure_ddp(
-        self, model: "LightningModule", device_ids: List[int]
+        self, model: LightningModule, device_ids: List[int]
     ) -> DistributedDataParallel:
         model = LightningDistributedDataParallel(
             model, device_ids=device_ids, find_unused_parameters=True
         )
         return model
 
-    def configure_sync_batchnorm(self, model: "LightningModule") -> "LightningModule":
+    def configure_sync_batchnorm(self, model: LightningModule) -> LightningModule:
         """
         Add global batchnorm for a model spread across multiple GPUs and nodes.
 

pytorch_lightning/accelerators/ddp_cpu_slurm_accelerator.py

@@ -12,19 +12,20 @@
 # See the License for the specific language governing permissions and
 # limitations under the License
 import os
+from typing import List, Optional
+
 import torch
 import torch.distributed as torch_distrib
 import torch.distributed as dist
+from torch.nn.parallel import DistributedDataParallel
 
-from pytorch_lightning.accelerators.accelerator import Accelerator
 from pytorch_lightning import _logger as log
+from pytorch_lightning.accelerators.accelerator import Accelerator
+from pytorch_lightning.core.lightning import LightningModule
 from pytorch_lightning.utilities import AMPType
 from pytorch_lightning.utilities.distributed import rank_zero_only
-from pytorch_lightning.utilities.seed import seed_everything
 from pytorch_lightning.distributed.dist import LightningDistributed
 from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
-from torch.nn.parallel import DistributedDataParallel
-from typing import List, Optional
 
 
 try:
@@ -177,14 +178,14 @@ def ddp_train(self, process_idx, model):
         return results
 
     def configure_ddp(
-        self, model: "LightningModule", device_ids: List[int]
+        self, model: LightningModule, device_ids: List[int]
     ) -> DistributedDataParallel:
         model = LightningDistributedDataParallel(
             model, device_ids=device_ids, find_unused_parameters=True
         )
         return model
 
-    def configure_sync_batchnorm(self, model: "LightningModule") -> "LightningModule":
+    def configure_sync_batchnorm(self, model: LightningModule) -> LightningModule:
         """
         Add global batchnorm for a model spread across multiple GPUs and nodes.
 

pytorch_lightning/accelerators/ddp_cpu_spawn_accelerator.py

@@ -12,23 +12,22 @@
 # See the License for the specific language governing permissions and
 # limitations under the License
 import os
-import re
+from typing import List, Optional
 
 import torch
 import torch.distributed as torch_distrib
 import torch.distributed as dist
 import torch.multiprocessing as mp
+from torch.nn.parallel import DistributedDataParallel
 
 from pytorch_lightning import _logger as log
 from pytorch_lightning.accelerators.accelerator import Accelerator
+from pytorch_lightning.core.lightning import LightningModule
+from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
 from pytorch_lightning.utilities import AMPType
-from pytorch_lightning.utilities.cloud_io import atomic_save, load as pl_load
 from pytorch_lightning.utilities.distributed import rank_zero_only, rank_zero_warn
 from pytorch_lightning.utilities.distributed import find_free_network_port
 from pytorch_lightning.distributed.dist import LightningDistributed
-from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
-from torch.nn.parallel import DistributedDataParallel
-from typing import List, Optional
 
 try:
     from hydra.core.hydra_config import HydraConfig
@@ -210,14 +209,14 @@ def transfer_distrib_spawn_state_on_fit_end(self, model, mp_queue, results):
             mp_queue.put(results)
 
     def configure_ddp(
-        self, model: "LightningModule", device_ids: List[int]
+        self, model: LightningModule, device_ids: List[int]
     ) -> DistributedDataParallel:
         model = LightningDistributedDataParallel(
             model, device_ids=device_ids, find_unused_parameters=True
         )
         return model
 
-    def configure_sync_batchnorm(self, model: "LightningModule") -> "LightningModule":
+    def configure_sync_batchnorm(self, model: LightningModule) -> LightningModule:
         """
         Add global batchnorm for a model spread across multiple GPUs and nodes.
 

pytorch_lightning/accelerators/ddp_cpu_torchelastic_accelerator.py

@@ -12,19 +12,20 @@
 # See the License for the specific language governing permissions and
 # limitations under the License
 import os
+from typing import List, Optional
+
 import torch
 import torch.distributed as torch_distrib
 import torch.distributed as dist
+from torch.nn.parallel import DistributedDataParallel
 
-from pytorch_lightning.accelerators.accelerator import Accelerator
 from pytorch_lightning import _logger as log
-from pytorch_lightning.utilities import AMPType
-from pytorch_lightning.utilities.distributed import rank_zero_only
-from pytorch_lightning.utilities.seed import seed_everything
+from pytorch_lightning.accelerators.accelerator import Accelerator
+from pytorch_lightning.core.lightning import LightningModule
 from pytorch_lightning.distributed.dist import LightningDistributed
 from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
-from torch.nn.parallel import DistributedDataParallel
-from typing import List, Optional
+from pytorch_lightning.utilities import AMPType
+from pytorch_lightning.utilities.distributed import rank_zero_only
 
 try:
     from hydra.utils import to_absolute_path, get_original_cwd
@@ -176,14 +177,14 @@ def ddp_train(self, process_idx, model):
         return results
 
     def configure_ddp(
-        self, model: "LightningModule", device_ids: List[int]
+        self, model: LightningModule, device_ids: List[int]
     ) -> DistributedDataParallel:
         model = LightningDistributedDataParallel(
             model, device_ids=device_ids, find_unused_parameters=True
         )
         return model
 
-    def configure_sync_batchnorm(self, model: "LightningModule") -> "LightningModule":
+    def configure_sync_batchnorm(self, model: LightningModule) -> LightningModule:
         """
         Add global batchnorm for a model spread across multiple GPUs and nodes.
 

pytorch_lightning/accelerators/ddp_slurm_accelerator.py

@@ -12,19 +12,21 @@
 # See the License for the specific language governing permissions and
 # limitations under the License
 import os
+from typing import List
+
 import torch
 import torch.distributed as torch_distrib
 import torch.distributed as dist
+from torch.nn.parallel import DistributedDataParallel
 
-from pytorch_lightning.accelerators.accelerator import Accelerator
 from pytorch_lightning import _logger as log
+from pytorch_lightning.accelerators.accelerator import Accelerator
+from pytorch_lightning.core.lightning import LightningModule
+from pytorch_lightning.distributed.dist import LightningDistributed
+from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
 from pytorch_lightning.utilities import AMPType
 from pytorch_lightning.utilities.distributed import rank_zero_only
 from pytorch_lightning.utilities.seed import seed_everything
-from pytorch_lightning.distributed.dist import LightningDistributed
-from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
-from torch.nn.parallel import DistributedDataParallel
-from typing import List
 
 try:
     from hydra.utils import to_absolute_path, get_original_cwd
@@ -182,14 +184,14 @@ def ddp_train(self, process_idx, model):
         return results
 
     def configure_ddp(
-        self, model: "LightningModule", device_ids: List[int]
+        self, model: LightningModule, device_ids: List[int]
     ) -> DistributedDataParallel:
         model = LightningDistributedDataParallel(
             model, device_ids=device_ids, find_unused_parameters=True
         )
         return model
 
-    def configure_sync_batchnorm(self, model: "LightningModule") -> "LightningModule":
+    def configure_sync_batchnorm(self, model: LightningModule) -> LightningModule:
         """
         Add global batchnorm for a model spread across multiple GPUs and nodes.
 

pytorch_lightning/accelerators/ddp_spawn_accelerator.py

@@ -13,23 +13,23 @@
 # limitations under the License
 import os
 import re
+from typing import List, Optional
 
 import torch
 import torch.multiprocessing as mp
 import torch.distributed as torch_distrib
 import torch.distributed as dist
+from torch.nn.parallel import DistributedDataParallel
 
 from pytorch_lightning import _logger as log
 from pytorch_lightning.accelerators.accelerator import Accelerator
+from pytorch_lightning.core.lightning import LightningModule
+from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
 from pytorch_lightning.utilities import AMPType
 from pytorch_lightning.utilities.cloud_io import atomic_save, load as pl_load
-from pytorch_lightning.utilities.distributed import rank_zero_only, rank_zero_warn
+from pytorch_lightning.utilities.distributed import rank_zero_only, rank_zero_warn, find_free_network_port
 from pytorch_lightning.utilities.seed import seed_everything
 from pytorch_lightning.distributed.dist import LightningDistributed
-from pytorch_lightning.utilities.distributed import find_free_network_port
-from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
-from torch.nn.parallel import DistributedDataParallel
-from typing import List, Optional
 
 try:
     from hydra.core.hydra_config import HydraConfig
@@ -237,14 +237,14 @@ def transfer_distrib_spawn_state_on_fit_end(self, model, mp_queue, results):
             mp_queue.put(last_path)
 
     def configure_ddp(
-        self, model: "LightningModule", device_ids: List[int]
+        self, model: LightningModule, device_ids: List[int]
     ) -> DistributedDataParallel:
         model = LightningDistributedDataParallel(
             model, device_ids=device_ids, find_unused_parameters=True
         )
         return model
 
-    def configure_sync_batchnorm(self, model: "LightningModule") -> "LightningModule":
+    def configure_sync_batchnorm(self, model: LightningModule) -> LightningModule:
         """
         Add global batchnorm for a model spread across multiple GPUs and nodes.
 

pytorch_lightning/accelerators/ddp_torchelastic_accelerator.py

@@ -12,19 +12,20 @@
 # See the License for the specific language governing permissions and
 # limitations under the License
 import os
+from typing import List, Optional
+
 import torch
 import torch.distributed as torch_distrib
 import torch.distributed as dist
+from torch.nn.parallel import DistributedDataParallel
 
-from pytorch_lightning.accelerators.accelerator import Accelerator
 from pytorch_lightning import _logger as log
-from pytorch_lightning.utilities import AMPType
-from pytorch_lightning.utilities.distributed import rank_zero_only
-from pytorch_lightning.utilities.seed import seed_everything
+from pytorch_lightning.accelerators.accelerator import Accelerator
+from pytorch_lightning.core.lightning import LightningModule
 from pytorch_lightning.distributed.dist import LightningDistributed
 from pytorch_lightning.overrides.data_parallel import LightningDistributedDataParallel
-from torch.nn.parallel import DistributedDataParallel
-from typing import List, Optional
+from pytorch_lightning.utilities import AMPType
+from pytorch_lightning.utilities.distributed import rank_zero_only
 
 
 try:
@@ -179,14 +180,14 @@ def ddp_train(self, process_idx, model):
         return results
 
     def configure_ddp(
-        self, model: "LightningModule", device_ids: List[int]
+        self, model: LightningModule, device_ids: List[int]
     ) -> DistributedDataParallel:
         model = LightningDistributedDataParallel(
             model, device_ids=device_ids, find_unused_parameters=True
         )
         return model
 
-    def configure_sync_batchnorm(self, model: "LightningModule") -> "LightningModule":
+    def configure_sync_batchnorm(self, model: LightningModule) -> LightningModule:
         """
         Add global batchnorm for a model spread across multiple GPUs and nodes.
 

pytorch_lightning/accelerators/horovod_accelerator.py

@@ -152,7 +152,7 @@ def test_step(self, args):
         return output
 
     def backward(self, closure_loss, optimizer, opt_idx, *args, **kwargs):
-        super().backward(closure_loss, optimizer, opt_idx,  *args, **kwargs)
+        super().backward(closure_loss, optimizer, opt_idx, *args, **kwargs)
         optimizer.synchronize()
 
     def on_train_epoch_end(self, outputs):

pytorch_lightning/callbacks/early_stopping.py

@@ -35,8 +35,6 @@
 torch_inf = torch.tensor(np.Inf)
 
 
-
-
 class EarlyStopping(Callback):
     r"""
     Monitor a validation metric and stop training when it stops improving.

pytorch_lightning/core/__init__.py

@@ -11,6 +11,7 @@
 # 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.core.datamodule import LightningDataModule
 from pytorch_lightning.core.lightning import LightningModule