diff --git a/skada/deep/tests/test_deep_scorer.py b/skada/deep/tests/test_deep_scorer.py
index ae9913e4..30518b5e 100644
--- a/skada/deep/tests/test_deep_scorer.py
+++ b/skada/deep/tests/test_deep_scorer.py
@@ -14,6 +14,7 @@
 from skada.metrics import (
     CircularValidation,
     DeepEmbeddedValidation,
+    ImportanceWeightedScorer,
     MixValScorer,
     PredictionEntropyScorer,
     SoftNeighborhoodDensity,
@@ -28,6 +29,7 @@
         SoftNeighborhoodDensity(),
         CircularValidation(),
         MixValScorer(),
+        ImportanceWeightedScorer(),
     ],
 )
 def test_generic_scorer_on_deepmodel(scorer, da_dataset):
@@ -93,12 +95,18 @@ def test_generic_scorer(scorer, da_dataset):
     assert np.all(~np.isnan(scores)), "all scores are computed"
 
 
-def test_dev_cnn_features_nd(da_dataset):
+@pytest.mark.parametrize(
+    "scorer",
+    [
+        DeepEmbeddedValidation(),
+        ImportanceWeightedScorer(),
+    ],
+)
+def test_scorer_with_nd_features(scorer, da_dataset):
     X, y, sample_domain = da_dataset.pack_train(as_sources=["s"], as_targets=["t"])
     X = np.repeat(X[..., np.newaxis], repeats=5, axis=-1)  # Make it batched 2D data
     X = X.astype(np.float32)
 
-    scorer = DeepEmbeddedValidation()
     _, n_channels, input_size = X.shape
     y_source, _ = source_target_split(y, sample_domain=sample_domain)
     n_classes = len(np.unique(y_source))
diff --git a/skada/metrics.py b/skada/metrics.py
index 48f9d7d0..307e9114 100644
--- a/skada/metrics.py
+++ b/skada/metrics.py
@@ -153,8 +153,6 @@ def _fit(self, X_source, X_target):
             weight_estimator = KernelDensity()
         self.weight_estimator_source_ = clone(weight_estimator)
         self.weight_estimator_target_ = clone(weight_estimator)
-        X_source = X_source.reshape(X_source.shape[0], -1)
-        X_target = X_target.reshape(X_target.shape[0], -1)
         self.weight_estimator_source_.fit(X_source)
         self.weight_estimator_target_.fit(X_target)
         return self
@@ -173,11 +171,14 @@ def _score(self, estimator, X, y, sample_domain=None, **params):
         X_source, X_target, y_source, _ = source_target_split(
             X, y, sample_domain=sample_domain
         )
-        X_source = X_source.reshape(X_source.shape[0], -1)
-        X_target = X_target.reshape(X_target.shape[0], -1)
-        self._fit(X_source, X_target)
-        ws = self.weight_estimator_source_.score_samples(X_source)
-        wt = self.weight_estimator_target_.score_samples(X_source)
+
+        # Reshape to 2D vectors
+        X_source_reshaped = X_source.reshape(X_source.shape[0], -1)
+        X_target_reshaped = X_target.reshape(X_target.shape[0], -1)
+
+        self._fit(X_source_reshaped, X_target_reshaped)
+        ws = self.weight_estimator_source_.score_samples(X_source_reshaped)
+        wt = self.weight_estimator_target_.score_samples(X_source_reshaped)
         weights = np.exp(wt - ws)
 
         if weights.sum() != 0:
@@ -186,15 +187,39 @@ def _score(self, estimator, X, y, sample_domain=None, **params):
             warnings.warn("All weights are zero. Using uniform weights.")
             weights = np.ones_like(weights) / len(weights)
 
-        return self._sign * scorer(
-            estimator,
-            X_source,
-            y_source,
-            sample_domain=sample_domain[sample_domain >= 0],
-            sample_weight=weights,
-            allow_source=True,
-            **params,
-        )
+        if isinstance(estimator, BaseEstimator):
+            score = scorer(
+                estimator,
+                X_source,
+                y_source,
+                sample_domain=sample_domain[sample_domain >= 0],
+                sample_weight=weights,
+                allow_source=True,
+                **params,
+            )
+        else:
+            try:
+                from skorch import NeuralNet
+
+                if isinstance(estimator, NeuralNet):
+                    # Deep estimators dont accept allow_source parameter
+                    score = scorer(
+                        estimator,
+                        X_source,
+                        y_source,
+                        sample_domain=sample_domain[sample_domain >= 0],
+                        sample_weight=weights,
+                        **params,
+                    )
+                else:
+                    raise ValueError("Estimator is not a NeuralNet instance")
+            except ImportError:
+                raise ValueError(
+                    "Importance Weighted Scorer does not support estimator"
+                    f"of type {type(estimator).__name__}"
+                )
+
+        return self._sign * score
 
 
 class PredictionEntropyScorer(_BaseDomainAwareScorer):