consistent verb tenses for method docstrings (ray-project#38108)

Signed-off-by: angelinalg <122562471+angelinalg@users.noreply.github.com>
Signed-off-by: e428265 <arvind.chandramouli@lmco.com>

python/ray/train/base_trainer.py

@@ -463,7 +463,7 @@ def _validate_attributes(self):
 
     @classmethod
     def _validate_scaling_config(cls, scaling_config: ScalingConfig) -> ScalingConfig:
-        """Return scaling config dataclass after validating updated keys."""
+        """Returns scaling config dataclass after validating updated keys."""
         ensure_only_allowed_dataclass_keys_updated(
             dataclass=scaling_config,
             allowed_keys=cls._scaling_config_allowed_keys,
@@ -472,7 +472,7 @@ def _validate_scaling_config(cls, scaling_config: ScalingConfig) -> ScalingConfi
 
     @classmethod
     def _maybe_sync_down_trainer_state(cls, restore_path: str) -> Path:
-        """Sync down trainer state from remote storage.
+        """Syncs down trainer state from remote storage.
 
         Returns:
             str: Local directory containing the trainer state
@@ -682,7 +682,7 @@ def _extract_fields_for_tuner_param_space(self) -> Dict:
         return result
 
     def _generate_trainable_cls(self) -> Type["Trainable"]:
-        """Generate the base Trainable class.
+        """Generates the base Trainable class.
 
         Returns:
             A Trainable class to use for training.
@@ -732,7 +732,7 @@ def train_func(config):
             dataset_context = None
 
         class TrainTrainable(trainable_cls):
-            """Add default resources to the Trainable."""
+            """Adds default resources to the Trainable."""
 
             _handles_checkpoint_freq = trainer_cls._handles_checkpoint_freq
             _handles_checkpoint_at_end = trainer_cls._handles_checkpoint_at_end
@@ -826,7 +826,7 @@ def default_resource_request(cls, config):
         return TrainTrainable
 
     def as_trainable(self) -> Type["Trainable"]:
-        """Convert self to a ``tune.Trainable`` class."""
+        """Converts self to a ``tune.Trainable`` class."""
         from ray import tune
 
         base_config = self._param_dict

python/ray/train/data_parallel_trainer.py

@@ -477,7 +477,7 @@ def training_loop(self) -> None:
         )
 
         def clear_lazy_checkpoint_marker():
-            """Clear the stale lazy checkpointing marker on all worker nodes.
+            """Clears the stale lazy checkpointing marker on all worker nodes.
 
             After recovery, the trainer may be scheduled on another node.
             We should delete the marker files created earlier on each node to
@@ -538,7 +538,7 @@ def clear_lazy_checkpoint_marker():
         backend_executor.shutdown()
 
     def get_dataset_config(self) -> DataConfig:
-        """Return a copy of this Trainer's final dataset configs.
+        """Returns a copy of this Trainer's final dataset configs.
 
         Returns:
             The merged default + user-supplied dataset config.
@@ -550,7 +550,7 @@ def get_dataset_config(self) -> DataConfig:
 
     @repr_with_fallback(["ipywidgets", "8"])
     def _repr_mimebundle_(self, **kwargs):
-        """Return a mimebundle with an ipywidget repr and a simple text repr.
+        """Returns a mimebundle with an ipywidget repr and a simple text repr.
 
         Depending on the frontend where the data is being displayed,
         different mimetypes will be used from this bundle.
@@ -656,7 +656,7 @@ def _datasets_repr_(self) -> str:
 def _load_checkpoint_dict(
     checkpoint: Checkpoint, trainer_name: str
 ) -> Tuple[Any, Optional["Preprocessor"]]:
-    """Load a Ray Train Checkpoint (dict based).
+    """Loads a Ray Train Checkpoint (dict based).
 
     This is a private API.