syntax

avalanche/benchmarks/classic/classic_benchmarks_utils.py

@@ -4,7 +4,9 @@
     DatasetStream,
 )
 
-from avalanche.benchmarks.utils.classification_dataset import TaskAwareClassificationDataset
+from avalanche.benchmarks.utils.classification_dataset import (
+    TaskAwareClassificationDataset,
+)
 
 
 def check_vision_benchmark(

avalanche/benchmarks/classic/ctrl.py

@@ -105,10 +105,12 @@ def CTrL(
                 paths_dataset: PathsDataset[Image, int] = PathsDataset(
                     common_root, exp_paths_list
                 )
-                dataset: TaskAwareSupervisedClassificationDataset = _make_taskaware_classification_dataset(
-                    paths_dataset,
-                    task_labels=task_labels,
-                    transform=transforms.Compose([transforms.ToTensor(), trans]),
+                dataset: TaskAwareSupervisedClassificationDataset = (
+                    _make_taskaware_classification_dataset(
+                        paths_dataset,
+                        task_labels=task_labels,
+                        transform=transforms.Compose([transforms.ToTensor(), trans]),
+                    )
                 )
             else:
                 dataset = _make_taskaware_tensor_classification_dataset(

avalanche/benchmarks/classic/endless_cl_sim.py

@@ -130,10 +130,14 @@ def EndlessCLSim(
         eval_data.transform = eval_transform
 
         train_datasets.append(
-            _make_taskaware_classification_dataset(dataset=train_data, task_labels=task_order[i])
+            _make_taskaware_classification_dataset(
+                dataset=train_data, task_labels=task_order[i]
+            )
         )
         eval_datasets.append(
-            _make_taskaware_classification_dataset(dataset=eval_data, task_labels=task_order[i])
+            _make_taskaware_classification_dataset(
+                dataset=eval_data, task_labels=task_order[i]
+            )
         )
 
     scenario_obj = dataset_benchmark(train_datasets, eval_datasets)

avalanche/benchmarks/scenarios/dataset_scenario.py

@@ -25,7 +25,8 @@
     Union,
     Tuple,
     Optional,
-    Iterable, Dict,
+    Iterable,
+    Dict,
 )
 
 from .generic_scenario import EagerCLStream, CLScenario, CLExperience, make_stream
@@ -35,9 +36,7 @@
 TCLDataset = TypeVar("TCLDataset", bound="AvalancheDataset")
 
 
-def benchmark_from_datasets(
-    **dataset_streams: Sequence[TCLDataset]
-) -> CLScenario:
+def benchmark_from_datasets(**dataset_streams: Sequence[TCLDataset]) -> CLScenario:
     """Creates a benchmark given a list of datasets for each stream.
 
     Each dataset will be considered as a separate experience.
@@ -59,7 +58,10 @@ def benchmark_from_datasets(
         for dd in data_s:
             if not isinstance(dd, AvalancheDataset):
                 raise ValueError("datasets must be AvalancheDatasets")
-        des = [DatasetExperience(dataset=dd, current_experience=eid) for eid, dd in enumerate(data_s)]
+        des = [
+            DatasetExperience(dataset=dd, current_experience=eid)
+            for eid, dd in enumerate(data_s)
+        ]
         s = EagerCLStream(stream_name, des)
         exps_streams.append(s)
     return CLScenario(exps_streams)
@@ -69,10 +71,7 @@ class DatasetExperience(CLExperience, Generic[TCLDataset]):
     """An Experience that provides a dataset."""
 
     def __init__(
-        self,
-        *,
-        dataset: TCLDataset,
-        current_experience: Optional[int] = None
+        self, *, dataset: TCLDataset, current_experience: Optional[int] = None
     ):
         super().__init__(current_experience=current_experience, origin_stream=None)
         self._dataset: AvalancheDataset = dataset
@@ -84,7 +83,9 @@ def dataset(self) -> AvalancheDataset:
         return data
 
 
-def _split_dataset_by_attribute(data: AvalancheDataset, attr_name: str) -> Dict[int, AvalancheDataset]:
+def _split_dataset_by_attribute(
+    data: AvalancheDataset, attr_name: str
+) -> Dict[int, AvalancheDataset]:
     """Helper to split a dataset by attribute.
 
     :param data: an Avalanche dataset.
@@ -101,8 +102,8 @@ def _split_dataset_by_attribute(data: AvalancheDataset, attr_name: str) -> Dict[
 def split_validation_random(
     validation_size: Union[int, float],
     shuffle: bool,
-    seed: int=None,
-    dataset: AvalancheDataset=None,
+    seed: int = None,
+    dataset: AvalancheDataset = None,
 ) -> Tuple[AvalancheDataset, AvalancheDataset]:
     """Splits an `AvalancheDataset` in two splits.
 
@@ -170,7 +171,9 @@ def split_validation_random(
 def split_validation_class_balanced(
     validation_size: Union[int, float],
     dataset: TaskAwareSupervisedClassificationDataset,
-) -> Tuple[TaskAwareSupervisedClassificationDataset, TaskAwareSupervisedClassificationDataset]:
+) -> Tuple[
+    TaskAwareSupervisedClassificationDataset, TaskAwareSupervisedClassificationDataset
+]:
     """Class-balanced dataset split.
 
     This splitting strategy splits `dataset` into train and validation data of
@@ -223,7 +226,8 @@ def split_validation_class_balanced(
 
 
 class LazyTrainValSplitter:
-    def __init__(self,
+    def __init__(
+        self,
         split_strategy: Callable[
             [AvalancheDataset],
             Tuple[AvalancheDataset, AvalancheDataset],
@@ -241,7 +245,7 @@ def __init__(self,
         """
         self.split_strategy = split_strategy
         self.experiences = experiences
-    
+
     def __iter__(self):
         for new_experience in self.experiences:
             yield self.split_strategy(new_experience.dataset)
@@ -300,6 +304,7 @@ def benchmark_with_validation_stream(
     if split_strategy is None:
         if seed is None:
             seed = random.randint(0, 1000000)
+
         # functools.partial is a more compact option
         # However, MyPy does not understand what a partial is -_-
         def random_validation_split_strategy_wrapper(data):
@@ -328,13 +333,15 @@ def random_validation_split_strategy_wrapper(data):
     other_streams = benchmark.streams
 
     del other_streams["train"]
-    return CLScenario(streams=[train_stream, valid_stream] + list(other_streams.values()))
+    return CLScenario(
+        streams=[train_stream, valid_stream] + list(other_streams.values())
+    )
 
 
 __all__ = [
     "_split_dataset_by_attribute",
     "benchmark_from_datasets",
     "DatasetExperience",
     "split_validation_random",
-    "benchmark_with_validation_stream"
+    "benchmark_with_validation_stream",
 ]

avalanche/benchmarks/scenarios/deprecated/__init__.py

@@ -4,7 +4,9 @@
 from .. import CLExperience, CLStream, CLScenario
 
 TCLDataset = TypeVar("TCLDataset", bound="AvalancheDataset")
-TDatasetExperience = TypeVar("TDatasetExperience", bound="DatasetExperience")  # Implementation, defined here
+TDatasetExperience = TypeVar(
+    "TDatasetExperience", bound="DatasetExperience"
+)  # Implementation, defined here
 
 
 class DatasetExperience(CLExperience, Generic[TCLDataset], ABC):

avalanche/benchmarks/scenarios/deprecated/generators.py

@@ -49,7 +49,9 @@
     LazyDatasetSequence,
 )
 from avalanche.benchmarks.scenarios.deprecated.new_classes.nc_scenario import NCScenario
-from avalanche.benchmarks.scenarios.deprecated.new_instances.ni_scenario import NIScenario
+from avalanche.benchmarks.scenarios.deprecated.new_instances.ni_scenario import (
+    NIScenario,
+)
 from avalanche.benchmarks.scenarios.online import FixedSizeExperienceSplitter
 from avalanche.benchmarks.utils.classification_dataset import (
     SupportedDataset,
@@ -211,7 +213,9 @@ class "34" will be mapped to "1", class "11" to "2" and so on.
         train_dataset_sup = list(
             map(_as_taskaware_supervised_classification_dataset, train_dataset)
         )
-        test_dataset_sup = list(map(_as_taskaware_supervised_classification_dataset, test_dataset))
+        test_dataset_sup = list(
+            map(_as_taskaware_supervised_classification_dataset, test_dataset)
+        )
 
         (
             seq_train_dataset,
@@ -238,7 +242,9 @@ class "34" will be mapped to "1", class "11" to "2" and so on.
             # Overrides n_experiences (and per_experience_classes, already done)
             n_experiences = len(train_dataset)
     else:
-        seq_train_dataset = _as_taskaware_supervised_classification_dataset(train_dataset)
+        seq_train_dataset = _as_taskaware_supervised_classification_dataset(
+            train_dataset
+        )
         seq_test_dataset = _as_taskaware_supervised_classification_dataset(test_dataset)
 
     transform_groups = dict(train=(train_transform, None), eval=(eval_transform, None))
@@ -366,7 +372,9 @@ def ni_benchmark(
         train_dataset_sup = list(
             map(_as_taskaware_supervised_classification_dataset, train_dataset)
         )
-        test_dataset_sup = list(map(_as_taskaware_supervised_classification_dataset, test_dataset))
+        test_dataset_sup = list(
+            map(_as_taskaware_supervised_classification_dataset, test_dataset)
+        )
 
         (
             seq_train_dataset,
@@ -376,7 +384,9 @@ def ni_benchmark(
             train_dataset_sup, test_dataset_sup
         )
     else:
-        seq_train_dataset = _as_taskaware_supervised_classification_dataset(train_dataset)
+        seq_train_dataset = _as_taskaware_supervised_classification_dataset(
+            train_dataset
+        )
         seq_test_dataset = _as_taskaware_supervised_classification_dataset(test_dataset)
 
     transform_groups = dict(train=(train_transform, None), eval=(eval_transform, None))
@@ -588,11 +598,10 @@ def data_incremental_benchmark(
         # However, MyPy does not understand what a partial is -_-
         def split_strategy_wrapper(exp):
             ss = FixedSizeExperienceSplitter(
-                experience=exp, 
-                experience_size=experience_size, 
-                shuffle=shuffle
+                experience=exp, experience_size=experience_size, shuffle=shuffle
             )
             return [e.dataset for e in ss]
+
         split_strategy = split_strategy_wrapper
     else:
         split_strategy = custom_split_strategy

avalanche/benchmarks/scenarios/deprecated/generic_benchmark_creation.py

@@ -43,7 +43,9 @@
 from avalanche.benchmarks.utils.classification_dataset import (
     TaskAwareClassificationDataset,
 )
-from avalanche.benchmarks.scenarios.deprecated.classification_scenario import GenericCLScenario
+from avalanche.benchmarks.scenarios.deprecated.classification_scenario import (
+    GenericCLScenario,
+)
 
 
 def create_multi_dataset_generic_benchmark(
@@ -141,7 +143,9 @@ def create_multi_dataset_generic_benchmark(
                 "complete_test_set_only is True"
             )
 
-    stream_definitions: Dict[str, Tuple[Iterable[TaskAwareClassificationDataset]]] = dict()
+    stream_definitions: Dict[
+        str, Tuple[Iterable[TaskAwareClassificationDataset]]
+    ] = dict()
 
     for stream_name, dataset_list in input_streams.items():
         initial_transform_group = "train"

avalanche/benchmarks/scenarios/deprecated/new_classes/nc_scenario.py

@@ -18,7 +18,9 @@
     ClassificationStream,
     ClassificationExperience,
 )
-from avalanche.benchmarks.utils.classification_dataset import _taskaware_classification_subset
+from avalanche.benchmarks.utils.classification_dataset import (
+    _taskaware_classification_subset,
+)
 from avalanche.benchmarks.utils.classification_dataset import (
     TaskAwareClassificationDataset,
     TaskAwareSupervisedClassificationDataset,
@@ -28,7 +30,9 @@
 
 
 class NCScenario(
-    ClassificationScenario["NCStream", "NCExperience", TaskAwareSupervisedClassificationDataset]
+    ClassificationScenario[
+        "NCStream", "NCExperience", TaskAwareSupervisedClassificationDataset
+    ]
 ):
 
     """

avalanche/benchmarks/scenarios/deprecated/new_instances/ni_scenario.py

@@ -30,7 +30,9 @@
 
 
 class NIScenario(
-    ClassificationScenario["NIStream", "NIExperience", TaskAwareSupervisedClassificationDataset]
+    ClassificationScenario[
+        "NIStream", "NIExperience", TaskAwareSupervisedClassificationDataset
+    ]
 ):
     """
     This class defines a "New Instance" scenario.
@@ -168,7 +170,9 @@ def __init__(
             included_patterns: List[int] = list()
             for exp_def in lst_fixed_exp_assignment:
                 included_patterns.extend(exp_def)
-            subset = _taskaware_classification_subset(train_dataset, indices=included_patterns)
+            subset = _taskaware_classification_subset(
+                train_dataset, indices=included_patterns
+            )
             unique_targets, unique_count = torch.unique(
                 torch.as_tensor(subset.targets), return_counts=True
             )

avalanche/benchmarks/scenarios/generic_scenario.py

@@ -620,7 +620,7 @@ def make_stream(name: str, exps: Iterable[CLExperience]) -> CLStream:
     s_wrapped: CLStream
     if isinstance(exps, List):  # Maintain indexing/slicing functionalities
         return EagerCLStream(name=name, exps=exps)
-    elif hasattr(exps, '__len__'):  # Sized stream
+    elif hasattr(exps, "__len__"):  # Sized stream
         return SizedCLStream(name=name, exps_iter=exps)
     else:  # Plain iterator
         return CLStream(name=name, exps_iter=exps)
@@ -635,5 +635,5 @@ def make_stream(name: str, exps: Iterable[CLExperience]) -> CLStream:
     "SequenceCLStream",
     "EagerCLStream",
     "CLScenario",
-    "make_stream"
+    "make_stream",
 ]

avalanche/benchmarks/scenarios/online.py

@@ -18,7 +18,8 @@
     List,
     Optional,
     TypeVar,
-    Union, Protocol,
+    Union,
+    Protocol,
 )
 from typing_extensions import Literal
 import warnings
@@ -110,7 +111,7 @@ def __init__(
         :type sub_stream_length: int, optional
         :param access_task_boundaries: Whether to access task boundaries.
         :type access_task_boundaries: bool, optional
-        """       
+        """
         super().__init__(dataset=dataset)
         self.access_task_boundaries = access_task_boundaries
         self.origin_experience: DatasetExperience = origin_experience
@@ -146,7 +147,7 @@ def __init__(
 
         Experience decorators (e.g. class attributes) will be stripped from the experience.
         You will need to re-apply them to the resulting experiences if you need them.
-        
+
         :param experience: The experience to split.
         :param experience_size: The experience size (number of instances).
         :param shuffle: If True, instances will be shuffled before splitting.
@@ -163,7 +164,7 @@ def __init__(
 
         # we need to fix the seed because repeated calls to the generator
         # must return the same order every time.
-        self.seed = random.randint(0, 2 ** 32 - 1)
+        self.seed = random.randint(0, 2**32 - 1)
 
     def __iter__(self):
         exp_dataset = self.experience.dataset
@@ -272,7 +273,9 @@ def split_online_stream(
         # functools.partial is a more compact option
         # However, MyPy does not understand what a partial is -_-
         def default_online_split_wrapper(e, e_sz):
-            return _default_online_split(shuffle, drop_last, access_task_boundaries, e, e_sz)
+            return _default_online_split(
+                shuffle, drop_last, access_task_boundaries, e, e_sz
+            )
 
         split_strategy = default_online_split_wrapper
     else:
@@ -342,9 +345,11 @@ def __init__(
         :param shuffle: If True, experiences will be split by first shuffling
             instances in each experience. Defaults to True.
         """
-        warnings.warn("Deprecated. Use `split_online_stream` or similar methods to split"
-                     "single streams or experiences instead")
-
+        warnings.warn(
+            "Deprecated. Use `split_online_stream` or similar methods to split"
+            "single streams or experiences instead"
+        )
+
         if stream_split_strategy != "fixed_size_split":
             raise ValueError("Unknown experience split strategy")