Convert directory fbcode/torcheval to use the Ruff Formatter

Summary:
Converts the directory specified to use the Ruff formatter in pyfmt

ruff_dog

If this diff causes merge conflicts when rebasing, please run
`hg status -n -0 --change . -I '**/*.{py,pyi}' | xargs -0 arc pyfmt`
on your diff, and amend any changes before rebasing onto latest.
That should help reduce or eliminate any merge conflicts.

allow-large-files

Reviewed By: amyreese

Differential Revision: D66304552

fbshipit-source-id: c138069c4b2b26c6d0fb3d47496ea717f40c8674

docs/source/conf.py

@@ -112,9 +112,7 @@ def setup(app):
 
     # In Sphinx 1.8 it was renamed to `add_css_file`, 1.7 and prior it is
     # `add_stylesheet` (deprecated in 1.8).
-    add_css = getattr(
-        app, "add_css_file", getattr(app, "add_stylesheet", None)
-    )  # noqa B009
+    add_css = getattr(app, "add_css_file", getattr(app, "add_stylesheet", None))  # noqa B009
     for css_file in html_css_files:
         add_css(css_file)
     app.set_translator("html", PatchedHTMLTranslator)

tests/metrics/aggregation/test_auc.py

@@ -142,8 +142,9 @@ def test_auc_class_no_unsqueeze(self) -> None:
         ]
 
         # Build sklearn metrics
-        sk_x, sk_y = torch.cat([tx.unsqueeze(0) for tx in x], dim=1), torch.cat(
-            [ty.unsqueeze(0) for ty in y], dim=1
+        sk_x, sk_y = (
+            torch.cat([tx.unsqueeze(0) for tx in x], dim=1),
+            torch.cat([ty.unsqueeze(0) for ty in y], dim=1),
         )
         compute_result = torch.tensor(
             [sklearn_auc(temp_x, temp_y) for temp_x, temp_y in zip(sk_x, sk_y)],

tests/metrics/classification/test_accuracy.py

@@ -283,7 +283,6 @@ def test_accuracy_class_invalid_input(self) -> None:
 
 class TestTopKAccuracy(MetricClassTester):
     def test_accuracy_class_base(self) -> None:
-
         input = torch.tensor(
             [
                 [
@@ -315,7 +314,6 @@ def test_accuracy_class_base(self) -> None:
         )
 
     def test_accuracy_class_macro(self) -> None:
-
         input = torch.tensor(
             [
                 [
@@ -347,7 +345,6 @@ def test_accuracy_class_macro(self) -> None:
         )
 
     def test_accuracy_class_no_average(self) -> None:
-
         input = torch.tensor(
             [
                 [

tests/metrics/classification/test_binned_precision_recall_curve.py

@@ -36,7 +36,6 @@ def _test_binary_binned_precision_recall_curve_class_with_input(
         target: torch.Tensor,
         threshold: Union[int, List[float], torch.Tensor],
     ) -> None:
-
         compute_result = binary_binned_precision_recall_curve(
             input.reshape(-1), target.reshape(-1), threshold=threshold
         )

tests/metrics/classification/test_confusion_matrix.py

@@ -26,7 +26,6 @@ def _test_binary_confusion_matrix_with_input(
         target: torch.Tensor,
         normalize: Optional[str] = None,
     ) -> None:
-
         input_np = input.flatten().numpy()
         target_np = target.flatten().numpy()
 
@@ -145,7 +144,6 @@ def test_binary_confusion_matrix_score_thresholding(self) -> None:
         )
 
     def test_binary_confusion_matrix_invalid_input(self) -> None:
-
         metric = BinaryConfusionMatrix()
 
         with self.assertRaisesRegex(
@@ -189,7 +187,6 @@ def _test_multiclass_confusion_matrix_with_input(
         num_classes: int,
         normalize: Optional[str] = None,
     ) -> None:
-
         if input.ndim == 3:
             input_np = input.argmax(dim=2).flatten().numpy()
         else:

tests/metrics/classification/test_precision.py

@@ -28,7 +28,6 @@ def _test_binary_precision_with_input(
         target: torch.Tensor,
         threshold: float = 0.5,
     ) -> None:
-
         input_np = np.where(input.numpy() < threshold, 0, 1)
         target_np = target.squeeze().numpy()
         compute_result = torch.tensor(

tests/metrics/classification/test_recall.py

@@ -238,7 +238,6 @@ def _test_binary_recall_class_with_input(
         target: torch.Tensor,
         threshold: float = 0.5,
     ) -> None:
-
         input_np = np.where(input.numpy() < threshold, 0, 1)
         target_np = target.squeeze().numpy()
 

tests/metrics/functional/classification/test_accuracy.py

@@ -64,7 +64,6 @@ def test_binary_accuracy_with_rounding(self) -> None:
         self._test_binary_accuracy_with_input(input, target)
 
     def test_binary_accuracy_invalid_input(self) -> None:
-
         with self.assertRaisesRegex(
             ValueError,
             "The `input` and `target` should have the same dimensions, "

tests/metrics/functional/classification/test_auprc.py

@@ -23,7 +23,6 @@ class TestBinaryAUPRC(unittest.TestCase):
     def _get_sklearn_equivalent(
         self, input: torch.Tensor, target: torch.Tensor, device: str = "cpu"
     ) -> torch.Tensor:
-
         # Convert input/target to sklearn style inputs
         # run each task once at a time since no multi-task/multiclass
         # available for sklearn
@@ -43,7 +42,6 @@ def _test_binary_auprc_with_input(
         num_tasks: int = 1,
         compute_result: Optional[torch.Tensor] = None,
     ) -> None:
-
         device = "cpu"
         if torch.cuda.is_available():
             device = "cuda"
@@ -175,7 +173,6 @@ def _test_multiclass_auprc_with_input(
         num_classes: int,
         compute_result: Optional[torch.Tensor] = None,
     ) -> None:
-
         device = "cpu"
         if torch.cuda.is_available():
             device = "cuda"
@@ -372,7 +369,6 @@ def _test_multilabel_auprc_with_input(
         average: Optional[str] = "macro",
         compute_result: Optional[torch.Tensor] = None,
     ) -> None:
-
         device = "cuda" if torch.cuda.is_available() else "cpu"
 
         # get sklearn compute result if none given

tests/metrics/functional/classification/test_confusion_matrix.py

@@ -70,7 +70,6 @@ def test_binary_confusion_matrix_score_thresholding(self) -> None:
         )
 
     def test_binary_confusion_matrix_invalid_input(self) -> None:
-
         with self.assertRaisesRegex(
             ValueError,
             "input should be a one-dimensional tensor for binary confusion matrix, "

tests/metrics/functional/classification/test_f1_score.py

@@ -88,7 +88,6 @@ def test_binary_f1_score_thresholding(self) -> None:
         )
 
     def test_binary_f1_score_invalid_input(self) -> None:
-
         with self.assertRaisesRegex(
             ValueError,
             "input should be a one-dimensional tensor for binary f1 score, "

tests/metrics/functional/classification/test_precision.py

@@ -27,7 +27,6 @@ def _test_binary_precision_with_input(
         target: torch.Tensor,
         threshold: float = 0.5,
     ) -> None:
-
         input_np = np.where(input.numpy() < threshold, 0, 1)
         target_np = target.squeeze().numpy()
         sklearn_result = torch.tensor(precision_score(target_np, input_np)).to(
@@ -71,7 +70,6 @@ def test_binary_precision_with_rounding(self) -> None:
         self._test_binary_precision_with_input(input, target)
 
     def test_binary_precision_invalid_input(self) -> None:
-
         with self.assertRaisesRegex(
             ValueError,
             "The `input` and `target` should have the same dimensions, "

tests/metrics/functional/classification/test_recall.py

@@ -179,7 +179,6 @@ def _test_binary_recall_with_input(
         target: torch.Tensor,
         threshold: float = 0.5,
     ) -> None:
-
         my_compute_result = my_binary_recall_score(input, target, threshold=threshold)
 
         input_np = np.where(input.numpy() < threshold, 0, 1)

tests/metrics/functional/text/test_perplexity.py

@@ -14,7 +14,6 @@
 
 class Perplexity(unittest.TestCase):
     def test_perplexity(self) -> None:
-
         input = torch.tensor([[[0.3659, 0.7025, 0.3104]], [[0.0097, 0.6577, 0.1947]]])
         target = torch.tensor([[2], [1]])
         torch.testing.assert_close(
@@ -43,7 +42,6 @@ def test_perplexity(self) -> None:
         )
 
     def test_perplexity_with_ignore_index(self) -> None:
-
         input = torch.tensor([[[0.3659, 0.7025, 0.3104]], [[0.0097, 0.6577, 0.1947]]])
         target = torch.tensor([[2], [100]])
         torch.testing.assert_close(
@@ -72,7 +70,6 @@ def test_perplexity_with_ignore_index(self) -> None:
         )
 
     def test_perplexity_with_invalid_input(self) -> None:
-
         with self.assertRaisesRegex(
             ValueError, "target should be a two-dimensional tensor"
         ):

tests/metrics/functional/text/test_word_information_lost.py

@@ -14,7 +14,6 @@
 
 class TestWordInformationLost(unittest.TestCase):
     def test_word_information_lost(self) -> None:
-
         input = ["hello world", "welcome to the facebook"]
         target = ["hello metaverse", "welcome to meta"]
         torch.testing.assert_close(

tests/metrics/image/test_fid.py

@@ -63,7 +63,6 @@ def _get_random_data_FrechetInceptionDistance(
         height: int,
         width: int,
     ) -> torch.Tensor:
-
         imgs = torch.rand(
             size=(num_updates, batch_size, num_channels, height, width),
         )
@@ -107,7 +106,6 @@ def _test_fid(
         expected_result: torch.Tensor,
         model: Optional[torch.nn.Module] = None,
     ) -> None:
-
         # create an alternating list of boolean values to
         # simulate a sequence of alternating real and generated images
         real_or_gen = [idx % 2 == 0 for idx in range(NUM_TOTAL_UPDATES)]

tests/metrics/image/test_psnr.py

@@ -91,7 +91,6 @@ def test_psnr_class_invalid_input(self) -> None:
             metric.update(torch.rand(4, 3, 4, 4), torch.rand(4, 3, 4, 6))
 
     def test_psnr_class_invalid_data_range(self) -> None:
-
         with self.assertRaisesRegex(
             ValueError, "`data_range needs to be either `None` or `float`."
         ):

tests/metrics/image/test_ssim.py

@@ -37,7 +37,6 @@ def _get_input_data(
         height: int,
         width: int,
     ) -> Dict[str, Tensor]:
-
         images = {
             "images_1": torch.rand(
                 size=(num_updates, batch_size, num_channels, height, width)
@@ -52,7 +51,6 @@ def _get_input_data(
     def test_ssim(
         self,
     ) -> None:
-
         images = self._get_input_data(
             NUM_TOTAL_UPDATES, BATCH_SIZE, IMG_CHANNELS, IMG_HEIGHT, IMG_WIDTH
         )

tests/metrics/ranking/test_retrieval_precision.py

@@ -14,7 +14,6 @@
 
 
 class TestRetrievalPrecision(MetricClassTester):
-
     # test update and compute functions
 
     def test_retrieval_precision_single_updates_1_query_avg_none(self) -> None:

tests/metrics/test_metric.py

@@ -171,7 +171,6 @@ def test_reset_state_tensor_dict(self) -> None:
         )
 
     def test_reset_primitive_type(self) -> None:
-
         TDummyMetric = TypeVar("TDummyMetric")
 
         class DummyMetric(Metric[Tuple[float, int]]):

tests/metrics/text/test_perplexity.py

@@ -67,7 +67,6 @@ def test_perplexity_with_ignore_index(self) -> None:
         )
 
     def test_perplexity_with_invalid_input(self) -> None:
-
         metric = Perplexity()
         with self.assertRaisesRegex(
             ValueError, "target should be a two-dimensional tensor"

torcheval/metrics/functional/classification/accuracy.py

@@ -256,7 +256,6 @@ def _multiclass_accuracy_update(
     num_classes: Optional[int],
     k: int,
 ) -> Tuple[torch.Tensor, torch.Tensor]:
-
     _accuracy_update_input_check(input, target, num_classes, k)
 
     if k == 1:
@@ -353,7 +352,6 @@ def _binary_accuracy_update(
     target: torch.Tensor,
     threshold: float = 0.5,
 ) -> Tuple[torch.Tensor, torch.Tensor]:
-
     _binary_accuracy_update_input_check(input, target)
 
     input = torch.where(input < threshold, 0, 1)
@@ -417,7 +415,6 @@ def _multilabel_update(
     target: torch.Tensor,
     criteria: str = "exact_match",
 ) -> Tuple[torch.Tensor, torch.Tensor]:
-
     if criteria == "exact_match":
         num_correct = torch.all(input == target, dim=1).sum()
         num_total = torch.tensor(target.shape[0], device=target.device)

torcheval/metrics/functional/classification/confusion_matrix.py

@@ -169,7 +169,6 @@ def _binary_confusion_matrix_update(
     target: torch.Tensor,
     threshold: float = 0.5,
 ) -> torch.Tensor:
-
     _binary_confusion_matrix_update_input_check(input, target)
 
     input = torch.where(input < threshold, 0, 1)
@@ -200,7 +199,6 @@ def _confusion_matrix_compute(
     confusion_matrix: torch.Tensor,
     normalize: Optional[str],
 ) -> torch.Tensor:
-
     if normalize == "pred":
         return norm(confusion_matrix.to(torch.float), p=1, dim=0)
     elif normalize == "true":

torcheval/metrics/functional/classification/f1_score.py

@@ -124,7 +124,6 @@ def _binary_f1_score_update(
     target: torch.Tensor,
     threshold: float = 0.5,
 ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
-
     _binary_f1_score_update_input_check(input, target)
 
     input = torch.where(input < threshold, 0, 1)

torcheval/metrics/functional/classification/precision.py

@@ -147,7 +147,6 @@ def _precision_compute(
     num_label: torch.Tensor,
     average: Optional[str],
 ) -> torch.Tensor:
-
     if average in ("macro", "weighted"):
         # Ignore the class that has no samples in both `input` and `target`
         mask = (num_label != 0) | ((num_tp + num_fp) != 0)
@@ -225,7 +224,6 @@ def _binary_precision_update(
     target: torch.Tensor,
     threshold: float = 0.5,
 ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
-
     _binary_precision_update_input_check(input, target)
 
     input = torch.where(input < threshold, 0, 1)

torcheval/metrics/functional/classification/recall.py

@@ -68,7 +68,6 @@ def _binary_recall_compute(
     num_tp: torch.Tensor,
     num_true_labels: torch.Tensor,
 ) -> torch.Tensor:
-
     recall = num_tp / num_true_labels
 
     if torch.isnan(recall):

torcheval/metrics/functional/image/psnr.py

@@ -48,7 +48,6 @@ def peak_signal_noise_ratio(
 
 
 def _psnr_param_check(data_range: Optional[float]) -> None:
-
     # Check matching shapes
     if data_range is not None:
         if type(data_range) is not float:
@@ -58,7 +57,6 @@ def _psnr_param_check(data_range: Optional[float]) -> None:
 
 
 def _psnr_input_check(input: torch.Tensor, target: torch.Tensor) -> None:
-
     # Check matching shapes
     if input.shape != target.shape:
         raise ValueError(

torcheval/metrics/functional/text/perplexity.py

@@ -113,7 +113,6 @@ def _perplexity_compute(
     sum_log_probs: torch.Tensor,
     num_total: torch.Tensor,
 ) -> torch.Tensor:
-
     return torch.exp(sum_log_probs / num_total).double()
 
 
@@ -122,7 +121,6 @@ def _perplexity_input_check(
     target: torch.Tensor,
     ignore_index: Optional[int] = None,
 ) -> None:
-
     if target.ndim != 2:
         raise ValueError(
             f"target should be a two-dimensional tensor, got shape {target.shape}."

torcheval/metrics/functional/text/word_information_lost.py

@@ -30,8 +30,8 @@ def _wil_update(
         input = [input]
     if isinstance(target, str):
         target = [target]
-    assert len(input) == len(
-        target
+    assert (
+        len(input) == len(target)
     ), f"Arguments must contain the same number of strings, but got len(input)={len(input)} and len(target)={len(target)}"
     errors, max_total, target_total, input_total = _get_errors_and_totals(input, target)
     return errors - max_total, target_total, input_total

torcheval/metrics/synclib.py

@@ -153,7 +153,7 @@ def send_tensors(
 
 
 def metrics_traversal_order(
-    state_dict: Dict[str, Dict[str, TState]]
+    state_dict: Dict[str, Dict[str, TState]],
 ) -> List[Tuple[str, str]]:
     """
     Args: