fixing examples (#6600)

* try Azure

* -e

* path

Author: Borda

azure-pipelines.yml

@@ -113,12 +113,12 @@ jobs:
         python -m pytest benchmarks -v --maxfail=2 --durations=0
       displayName: 'Testing: benchmarks'
 
-    - bash: |
+    - script: |
+        set -e
         python -m pytest pl_examples -v --maxfail=2 --durations=0
         python setup.py install --user --quiet
         bash pl_examples/run_ddp-example.sh
-        cd pl_examples/basic_examples
-        bash submit_ddp_job.sh
-        bash submit_ddp2_job.sh
-        pip uninstall -y pytorch-lightning
+        # cd pl_examples/basic_examples
+        # bash submit_ddp_job.sh
+        # bash submit_ddp2_job.sh
       displayName: 'Examples'

pl_examples/basic_examples/submit_ddp2_job.sh

@@ -24,4 +24,4 @@ source activate $1
 # -------------------------
 
 # run script from above
-srun python3 image_classifier.py --accelerator 'ddp2' --gpus 2 --num_nodes 2
+srun python3 simple_image_classifier.py --accelerator 'ddp2' --gpus 2 --num_nodes 2 --max_epochs 5

pl_examples/basic_examples/submit_ddp_job.sh

@@ -24,4 +24,4 @@ source activate $1
 # -------------------------
 
 # run script from above
-srun python3 image_classifier.py --accelerator 'ddp' --gpus 2 --num_nodes 2
+srun python3 simple_image_classifier.py --accelerator 'ddp' --gpus 2 --num_nodes 2 --max_epochs 5

tests/__init__.py

@@ -19,8 +19,8 @@
 _TEST_ROOT = os.path.dirname(__file__)
 _PROJECT_ROOT = os.path.dirname(_TEST_ROOT)
 _TEMP_PATH = os.path.join(_PROJECT_ROOT, 'test_temp')
-DATASETS_PATH = os.path.join(_PROJECT_ROOT, 'Datasets')
-LEGACY_PATH = os.path.join(_PROJECT_ROOT, 'legacy')
+PATH_DATASETS = os.path.join(_PROJECT_ROOT, 'Datasets')
+PATH_LEGACY = os.path.join(_PROJECT_ROOT, 'legacy')
 
 # todo: this setting `PYTHONPATH` may not be used by other evns like Conda for import packages
 if _PROJECT_ROOT not in os.getenv('PYTHONPATH', ""):

tests/base/model_template.py

@@ -18,6 +18,7 @@
 import torch.nn.functional as F
 
 from pytorch_lightning.core.lightning import LightningModule
+from tests import PATH_DATASETS
 from tests.base.model_optimizers import ConfigureOptimizersPool
 from tests.base.model_test_dataloaders import TestDataloaderVariations
 from tests.base.model_test_epoch_ends import TestEpochEndVariations
@@ -28,7 +29,7 @@
 from tests.base.model_valid_dataloaders import ValDataloaderVariations
 from tests.base.model_valid_epoch_ends import ValidationEpochEndVariations
 from tests.base.model_valid_steps import ValidationStepVariations
-from tests.helpers.datasets import PATH_DATASETS, TrialMNIST
+from tests.helpers.datasets import TrialMNIST
 
 
 class EvalModelTemplate(

tests/checkpointing/test_legacy_checkpoints.py

@@ -18,9 +18,9 @@
 import pytest
 
 from pytorch_lightning import Trainer
-from tests import LEGACY_PATH
+from tests import PATH_LEGACY
 
-LEGACY_CHECKPOINTS_PATH = os.path.join(LEGACY_PATH, 'checkpoints')
+LEGACY_CHECKPOINTS_PATH = os.path.join(PATH_LEGACY, 'checkpoints')
 CHECKPOINT_EXTENSION = ".ckpt"
 
 

tests/helpers/advanced_models.py

@@ -20,6 +20,7 @@
 from torch.utils.data import DataLoader
 
 from pytorch_lightning.core.lightning import LightningModule
+from tests import PATH_DATASETS
 from tests.helpers.datasets import AverageDataset, MNIST, TrialMNIST
 
 
@@ -165,7 +166,7 @@ def configure_optimizers(self):
         return [opt_g, opt_d], []
 
     def train_dataloader(self):
-        return DataLoader(TrialMNIST(train=True, download=True), batch_size=16)
+        return DataLoader(TrialMNIST(root=PATH_DATASETS, train=True, download=True), batch_size=16)
 
 
 class ParityModuleRNN(LightningModule):
@@ -223,6 +224,7 @@ def configure_optimizers(self):
 
     def train_dataloader(self):
         return DataLoader(MNIST(
+            root=PATH_DATASETS,
             train=True,
             download=True,
         ), batch_size=128, num_workers=1)

tests/helpers/datasets.py

@@ -22,11 +22,6 @@
 from torch import Tensor
 from torch.utils.data import Dataset
 
-from tests import _PROJECT_ROOT
-
-#: local path to test datasets
-PATH_DATASETS = os.path.join(_PROJECT_ROOT, 'Datasets')
-
 
 class MNIST(Dataset):
     """
@@ -47,7 +42,7 @@ class MNIST(Dataset):
             downloaded again.
 
     Examples:
-        >>> dataset = MNIST(download=True)
+        >>> dataset = MNIST(".", download=True)
         >>> len(dataset)
         60000
         >>> torch.bincount(dataset.targets)
@@ -65,7 +60,7 @@ class MNIST(Dataset):
 
     def __init__(
         self,
-        root: str = PATH_DATASETS,
+        root: str,
         train: bool = True,
         normalize: tuple = (0.1307, 0.3081),
         download: bool = True,
@@ -152,7 +147,7 @@ class TrialMNIST(MNIST):
         kwargs: Same as MNIST
 
     Examples:
-        >>> dataset = TrialMNIST(download=True)
+        >>> dataset = TrialMNIST(".", download=True)
         >>> len(dataset)
         300
         >>> sorted(set([d.item() for d in dataset.targets]))
@@ -161,15 +156,15 @@ class TrialMNIST(MNIST):
         tensor([100, 100, 100])
     """
 
-    def __init__(self, num_samples: int = 100, digits: Optional[Sequence] = (0, 1, 2), **kwargs):
+    def __init__(self, root: str, num_samples: int = 100, digits: Optional[Sequence] = (0, 1, 2), **kwargs):
         # number of examples per class
         self.num_samples = num_samples
         # take just a subset of MNIST dataset
         self.digits = sorted(digits) if digits else list(range(10))
 
         self.cache_folder_name = f"digits-{'-'.join(str(d) for d in self.digits)}_nb-{self.num_samples}"
 
-        super().__init__(normalize=(0.5, 1.0), **kwargs)
+        super().__init__(root, normalize=(0.5, 1.0), **kwargs)
 
     @staticmethod
     def _prepare_subset(full_data: torch.Tensor, full_targets: torch.Tensor, num_samples: int, digits: Sequence):

tests/helpers/test_datasets.py

@@ -16,12 +16,17 @@
 import cloudpickle
 import pytest
 
+from tests import PATH_DATASETS
 from tests.helpers.datasets import AverageDataset, MNIST, TrialMNIST
 
 
-@pytest.mark.parametrize('dataset_cls', [MNIST, TrialMNIST, AverageDataset])
-def test_pickling_dataset_mnist(tmpdir, dataset_cls):
-    mnist = dataset_cls()
+@pytest.mark.parametrize('dataset_cls,args', [
+    (MNIST, dict(root=PATH_DATASETS)),
+    (TrialMNIST, dict(root=PATH_DATASETS)),
+    (AverageDataset, dict()),
+])
+def test_pickling_dataset_mnist(tmpdir, dataset_cls, args):
+    mnist = dataset_cls(**args)
 
     mnist_pickled = pickle.dumps(mnist)
     pickle.loads(mnist_pickled)