Add missing dtype check in PrecisionRecallCurve (#1457)

Co-authored-by: stancld <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Jirka Borovec <[email protected]>
(cherry picked from commit 94c49f3)

CHANGELOG.md

@@ -34,6 +34,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Fixed type checking on the `maximize` parameter at the initialization of `MetricTracker` ([#1428](https://github.com/Lightning-AI/metrics/issues/1428))
 
 
+- Fix dtype checking in `PrecisionRecallCurve` for `target` tensor ([#1457](https://github.com/Lightning-AI/metrics/pull/1457))
+
 ## [0.11.0] - 2022-11-30
 
 ### Added

src/torchmetrics/functional/classification/precision_recall_curve.py

@@ -132,6 +132,12 @@ def _binary_precision_recall_curve_tensor_validation(
     """
     _check_same_shape(preds, target)
 
+    if target.is_floating_point():
+        raise ValueError(
+            "Expected argument `target` to be an int or long tensor with ground truth labels"
+            f" but got tensor with dtype {target.dtype}"
+        )
+
     if not preds.is_floating_point():
         raise ValueError(
             "Expected argument `preds` to be an floating tensor with probability/logit scores,"
@@ -334,6 +340,10 @@ def _multiclass_precision_recall_curve_tensor_validation(
         raise ValueError(
             f"Expected `preds` to have one more dimension than `target` but got {preds.ndim} and {target.ndim}"
         )
+    if target.is_floating_point():
+        raise ValueError(
+            f"Expected argument `target` to be an int or long tensor, but got tensor with dtype {target.dtype}"
+        )
     if not preds.is_floating_point():
         raise ValueError(f"Expected `preds` to be a float tensor, but got {preds.dtype}")
     if preds.shape[1] != num_classes:

tests/unittests/classification/test_precision_recall_curve.py

@@ -132,6 +132,16 @@ def test_binary_precision_recall_curve_threshold_arg(self, input, threshold_fn):
             assert torch.allclose(r1, r2)
             assert torch.allclose(t1, t2)
 
+    def test_binary_error_on_wrong_dtypes(self, input):
+        """Test that error are raised on wrong dtype."""
+        preds, target = input
+
+        with pytest.raises(ValueError, match="Expected argument `target` to be an int or long tensor with ground.*"):
+            binary_precision_recall_curve(preds[0], target[0].to(torch.float32))
+
+        with pytest.raises(ValueError, match="Expected argument `preds` to be an floating tensor with probability.*"):
+            binary_precision_recall_curve(preds[0].long(), target[0])
+
 
 def _sk_precision_recall_curve_multiclass(preds, target, ignore_index=None):
     preds = np.moveaxis(preds.numpy(), 1, -1).reshape((-1, preds.shape[1]))
@@ -243,6 +253,16 @@ def test_multiclass_precision_recall_curve_threshold_arg(self, input, threshold_
                 assert torch.allclose(r1[i], r2[i])
                 assert torch.allclose(t1[i], t2)
 
+    def test_multiclass_error_on_wrong_dtypes(self, input):
+        """Test that error are raised on wrong dtype."""
+        preds, target = input
+
+        with pytest.raises(ValueError, match="Expected argument `target` to be an int or long tensor, but got.*"):
+            multiclass_precision_recall_curve(preds[0], target[0].to(torch.float32), num_classes=NUM_CLASSES)
+
+        with pytest.raises(ValueError, match="Expected `preds` to be a float tensor, but got.*"):
+            multiclass_precision_recall_curve(preds[0].long(), target[0], num_classes=NUM_CLASSES)
+
 
 def _sk_precision_recall_curve_multilabel(preds, target, ignore_index=None):
     precision, recall, thresholds = [], [], []
@@ -345,6 +365,16 @@ def test_multilabel_precision_recall_curve_threshold_arg(self, input, threshold_
                 assert torch.allclose(r1[i], r2[i])
                 assert torch.allclose(t1[i], t2)
 
+    def test_multilabel_error_on_wrong_dtypes(self, input):
+        """Test that error are raised on wrong dtype."""
+        preds, target = input
+
+        with pytest.raises(ValueError, match="Expected argument `target` to be an int or long tensor with ground.*"):
+            multilabel_precision_recall_curve(preds[0], target[0].to(torch.float32), num_labels=NUM_CLASSES)
+
+        with pytest.raises(ValueError, match="Expected argument `preds` to be an floating tensor with probability.*"):
+            multilabel_precision_recall_curve(preds[0].long(), target[0], num_labels=NUM_CLASSES)
+
 
 @pytest.mark.parametrize(
     "metric",