Fix codestyle issues

catalyst/callbacks/checkpoint.py

@@ -158,6 +158,7 @@ def _load_states_from_file_map(
     from files specified in ``load_map``.
 
     Arguments:
+        logdir: directory with logs
         runner: current runner
         load_map (Dict[str, str]): dict with mappings to load.
             Expected keys - ``'model'``, ``'criterion'``
@@ -485,6 +486,15 @@ def _save_checkpoint(
         """
         Saves checkpoints: full with model/criterion/optimizer/scheduler
         and truncated with model only.
+
+        Args:
+            runner: current runner.
+            checkpoint: data to save.
+            is_best: if ``True`` then also will be generated best checkpoint file.
+            is_last: if ``True`` then also will be generated last checkpoint file.
+
+        Returns:
+            path to saved checkpoint
         """
         logdir = Path(f"{self.logdir}/")
         suffix = f"{runner.stage_key}.{runner.stage_epoch_step}"

catalyst/callbacks/scheduler.py

@@ -368,6 +368,9 @@ def __init__(
             scale: learning rate increasing scale ("log" or "linear")
             num_steps:  number of batches to try, if None - whole loader would be used.
             optimizer_key: which optimizer key to use for learning rate scheduling
+
+        Raises:
+            NotImplementedError: if invalid scale value.
         """
         super().__init__(optimizer_key=optimizer_key)
 

catalyst/callbacks/soft_update.py

@@ -24,14 +24,17 @@ def __init__(
         """Init.
 
         Args:
-        target_model_key: key to the data inside `runner.model` to update
-        source_model_key: key to the source data inside `runner.model`
-        tau: smoothing parameter `target * (1.0 - tau) + source * tau`
-        scope (str): when the `target` should be updated
+            target_model_key: key to the data inside `runner.model` to update
+            source_model_key: key to the source data inside `runner.model`
+            tau: smoothing parameter `target * (1.0 - tau) + source * tau`
+            scope (str): when the `target` should be updated
                 ``"on_batch_end"``
                 ``"on_batch_start"``
                 ``"on_epoch_end"``
                 ``"on_epoch_start"``
+
+        Raises:
+            TypeError: if invalid scope
         """
         super().__init__(order=CallbackOrder.Metric)
         self.target_model_key = target_model_key

catalyst/contrib/data/reader.py

@@ -35,7 +35,10 @@ def __call__(self, element):
             element: elem in your dataset
 
         Returns:
-            Data object used for your neural network
+            Data object used for your neural network  # noqa: DAR202
+
+        Raises:
+            NotImplementedError: you should implement it
         """
         raise NotImplementedError("You cannot apply a transformation using `BaseReader`")
 
@@ -145,7 +148,6 @@ def __init__(self, transforms: List[IReader]):
         """
         Args:
             transforms: list of reader to compose
-            mixins: list of mixins to use
         """
         self.transforms = transforms
 

catalyst/contrib/datasets/cv/imagecar.py

@@ -69,6 +69,9 @@ def __init__(
             transforms: (callable, optional): A function/transform that
                 takes in an image and returns a transformed version.
 
+        Raises:
+            RuntimeError: If ``download is False`` and the dataset not found.
+
         Examples:
             >>> from catalyst.contrib.datasets import CarvanaOneCarDataset
             >>> dataset = CarvanaOneCarDataset(root='./',

catalyst/contrib/datasets/cv/misc.py

@@ -55,7 +55,7 @@ def __init__(self, root: str, train: bool = True, download: bool = False, **kwar
             download: if ``True``, downloads the dataset from
                 the internet and puts it in root directory. If dataset
                 is already downloaded, it is not downloaded again
-            **kwargs:
+            **kwargs: Keyword-arguments passed to ``super().__init__`` method.
         """
         # downlad dataset if needed
         if download and not os.path.exists(os.path.join(root, self.name)):

catalyst/contrib/datasets/mnist.py

@@ -82,6 +82,9 @@ def __init__(self, root, train=True, transform=None, target_transform=None, down
                 takes in an image and returns a transformed version.
             target_transform (callable, optional): A function/transform
                 that takes in the target and transforms it.
+
+        Raises:
+            RuntimeError: If ``download is False`` and the dataset not found.
         """
         if isinstance(root, torch._six.string_classes):
             root = os.path.expanduser(root)
@@ -215,6 +218,9 @@ class MnistMLDataset(MetricLearningTrainDataset, MNIST):
 
     def __init__(self, **kwargs):
         """
+        Args:
+            **kwargs: Keyword-arguments passed to ``super().__init__`` method.
+
         Raises:
             ValueError: if train argument is False (MnistMLDataset
                 should be used only for training)

catalyst/contrib/datasets/movielens.py

@@ -71,6 +71,9 @@ def __init__(self, root, train=True, download=False, min_rating=0.0):
                 is already downloaded, it is not downloaded again.
             min_rating (float, optional): Minimum rating to include in
                 the interaction matrix
+
+        Raises:
+            RuntimeError: If ``download is False`` and the dataset not found.
         """
         if isinstance(root, torch._six.string_classes):
             root = os.path.expanduser(root)
@@ -196,7 +199,7 @@ def _parse(self, data):
         Args:
             data: raw data of the dataset
 
-        Returns:
+        Yields:
             Generator iterator for parsed data
         """
         for line in data:

catalyst/contrib/models/mnist.py

@@ -12,6 +12,7 @@ def __init__(self, out_features: int, normalize: bool = True):
         """
         Args:
             out_features: size of the output tensor
+            normalize: boolean flag to add normalize layer
         """
         super().__init__()
         layers = [

catalyst/contrib/nn/criterion/supervised_contrastive.py

@@ -27,8 +27,10 @@ def __init__(self, tau: float, reduction: str = "mean", pos_aggregation="in") ->
                 ``"in"`` | ``"out"``.
                 ``"in"``: maximization of log(average positive exponentiate similarity)
                 ``"out"``: maximization of average positive similarity
+
         Raises:
             ValueError: if reduction is not mean, sum or none
+            ValueError: if positive aggregation is not in or out
         """
         super().__init__()
         self.tau = tau

catalyst/contrib/nn/optimizers/lamb.py

@@ -51,6 +51,9 @@ def __init__(
                 (default: 0)
             adam (bool, optional): always use trust ratio = 1, which turns
                 this into Adam. Useful for comparison purposes.
+
+        Raises:
+            ValueError: if invalid learning rate, epsilon value or betas.
         """
         if not 0.0 <= lr:
             raise ValueError(f"Invalid learning rate: {lr}")

catalyst/contrib/nn/optimizers/qhadamw.py

@@ -51,6 +51,10 @@ def __init__(
             weight_decay (float, optional): weight decay
                 (L2 regularization coefficient, times two)
                 (default: 0.0)
+
+        Raises:
+            ValueError: if invalid learning rate, epsilon value, betas or
+                weight_decay value.
         """
         if not 0.0 <= lr:
             raise ValueError(f"Invalid learning rate: {lr}")

catalyst/data/dataset/metric_learning.py

@@ -39,6 +39,9 @@ def __getitem__(self, item) -> Dict[str, torch.Tensor]:
         be boolean and indicate whether current object
         is in query or in gallery.
 
+        Args:
+            item: Item
+
         Raises:
             NotImplementedError: You should implement it  # noqa: DAR402
         """

catalyst/data/dataset/self_supervised.py

@@ -64,6 +64,9 @@ def __init__(
             transform_original: transforms which will be applied to save original in batch
             is_target: the flag for selection does dataset return (sample, target) or only sample
 
+        Raises:
+            ValueError: should be specified transform_left and transform_right simultaneously
+                or only transforms
         """
         super().__init__()
 

catalyst/data/loader.py

@@ -131,6 +131,9 @@ def __next__(self):
 
         Returns:
             next batch
+
+        Raises:
+            StopIteration: if iteration_index >= len(origin)
         """
         if self.iteration_index >= len(self.origin):
             raise StopIteration()

catalyst/data/sampler.py

@@ -206,8 +206,8 @@ def __init__(self, labels: List[int], mode: Union[str, int] = "downsampling"):
 
     def __iter__(self) -> Iterator[int]:
         """
-        Yields:
-            indices of stratified sample
+        Returns:
+            iterator of indices of stratified sample
         """
         indices = []
         for key in sorted(self.lbl2idx):
@@ -487,8 +487,8 @@ def _exp_scheduler(self) -> float:
 
     def __iter__(self) -> Iterator[int]:
         """
-        Yields:
-            indices of stratified sample
+        Returns:
+            iterator of indices of stratified sample
         """
         indices = []
         for key in sorted(self.label2idxes):
@@ -708,11 +708,11 @@ def __init__(
         Args:
             sampler: Sampler used for subsampling
             num_replicas (int, optional): Number of processes participating in
-              distributed training
+                distributed training
             rank (int, optional): Rank of the current process
-              within ``num_replicas``
+                within ``num_replicas``
             shuffle (bool, optional): If true (default),
-              sampler will shuffle the indices
+                sampler will shuffle the indices
         """
         super(DistributedSamplerWrapper, self).__init__(
             DatasetFromSampler(sampler), num_replicas=num_replicas, rank=rank, shuffle=shuffle

catalyst/data/sampler_inbatch.py

@@ -138,7 +138,8 @@ def _sample(self, *_: Tensor, labels: List[int]) -> TTripletsIds:
             labels: labels of the samples in the batch
             *_: note, that we ignore features argument
 
-        Returns: indices of triplets
+        Returns:
+            indices of triplets
         """
         num_labels = len(labels)
 

catalyst/loggers/mlflow.py

@@ -3,6 +3,8 @@
 
 import numpy as np
 
+from mlflow.tracking.fluent import ActiveRun
+
 from catalyst.core.logger import ILogger
 from catalyst.settings import SETTINGS
 from catalyst.typing import Directory, File, Union
@@ -11,9 +13,15 @@
     import mlflow
 
 
-def _get_or_start_run(run_name):
+def _get_or_start_run(run_name: Optional[str]) -> ActiveRun:
     """The function of MLflow. Gets the active run and gives it a name.
     If active run does not exist, starts a new one.
+
+    Args:
+        run_name: Name of the run
+
+    Returns:
+        ActiveRun
     """
     active_run = mlflow.active_run()
     if active_run:

catalyst/metrics/_classification.py

@@ -35,6 +35,9 @@ class StatisticsMetric(ICallbackBatchMetric):
         prefix: metric prefix
         suffix: metric suffix
 
+    Raises:
+        ValueError: if mode is incorrect
+
     Examples:
 
     .. code-block:: python
@@ -100,11 +103,7 @@ def __init__(
         prefix: Optional[str] = None,
         suffix: Optional[str] = None,
     ):
-        """Init params
-
-        Raises:
-            ValueError: if mode is incorrect
-        """
+        """Init params"""
         super().__init__(compute_on_call=compute_on_call, prefix=prefix, suffix=suffix)
         if mode == "binary":
             self.statistics_fn = get_binary_statistics

catalyst/metrics/_metric.py

@@ -57,6 +57,10 @@ def __call__(self, *args, **kwargs) -> Any:
         (`on_batch_end` event).
         Returns computed value if `compute_on_call=True`.
 
+        Args:
+            *args: Arguments passed to update method.
+            **kwargs: Keyword-arguments passed to update method.
+
         Returns:
             Any: computed value, it's better to return key-value.
         """