fix linting

Author: stes

tests/_util.py

@@ -20,8 +20,9 @@
 # limitations under the License.
 #
 import collections.abc as collections_abc
-import packaging.version
 import functools
+
+import packaging.version
 import pytest
 import sklearn.utils.estimator_checks
 import torch
@@ -66,12 +67,18 @@ def parametrize_slow(arg_names, fast_arguments, slow_arguments):
 def parametrize_with_checks_slow(fast_arguments, slow_arguments):
 
     # NOTE(stes): See https://github.com/AdaptiveMotorControlLab/CEBRA/issues/280, sklearn API changed in 1.6.
-    if packaging.version.parse(sklearn.__version__) <= packaging.version.parse("1.6"):
-        generate_checks = functools.partial(sklearn.utils.estimator_checks.check_estimator, generate_only=True)
+    if packaging.version.parse(
+            sklearn.__version__) <= packaging.version.parse("1.6"):
+        generate_checks = functools.partial(
+            sklearn.utils.estimator_checks.check_estimator, generate_only=True)
     else:
         generate_checks = sklearn.utils.estimator_checks.estimator_checks_generator
-    generate_params = lambda args: [next(generate_checks(arg)) for arg in args]
-    return parametrize_slow("estimator,check", generate_params(fast_arguments), generate_params(slow_arguments))
+
+    def _generate_params(args):
+        return [next(generate_checks(arg)) for arg in args]
+
+    return parametrize_slow("estimator,check", _generate_params(fast_arguments),
+                            _generate_params(slow_arguments))
 
 
 def parametrize_device(func):