Add missing error raising and test for adapt on Multissession (#114)

Add missing error raising and test

Author: CeliaBenquet

cebra/integrations/sklearn/cebra.py

@@ -935,7 +935,7 @@ def _adapt_model(
 
         dataset, is_multisession = self._prepare_data(X, y)
 
-        if is_multisession:
+        if is_multisession or isinstance(self.model_, nn.ModuleList):
             raise NotImplementedError(
                 "The adapt option with a multisession training is not handled. Please use adapt=True for single-trained estimators only."
             )

tests/test_sklearn.py

@@ -675,6 +675,48 @@ def check_first_layer_dim(model, X):
         cebra_model.fit([X, X_s2], [y_c1, y_c1_s2], adapt=True)
 
 
+@_util.parametrize_slow(
+    arg_names="model_architecture,device",
+    fast_arguments=list(
+        itertools.islice(
+            itertools.product(
+                cebra_sklearn_cebra.CEBRA.supported_model_architectures(),
+                _DEVICES),
+            2,
+        )),
+    slow_arguments=list(
+        itertools.product(
+            cebra_sklearn_cebra.CEBRA.supported_model_architectures(),
+            _DEVICES)),
+)
+def test_sklearn_adapt_multisession(model_architecture, device):
+    num_hidden_units = 32
+    cebra_model = cebra_sklearn_cebra.CEBRA(
+        model_architecture=model_architecture,
+        time_offsets=10,
+        learning_rate=3e-4,
+        max_iterations=5,
+        max_adapt_iterations=1,
+        device=device,
+        output_dimension=4,
+        num_hidden_units=num_hidden_units,
+        batch_size=42,
+        verbose=True,
+    )
+    
+    # example dataset
+    Xs = [np.random.uniform(0, 1, (1000, 50)) for i in range(3)]
+    ys = [np.random.uniform(0, 1, (1000, 5)) for i in range(3)]
+    
+    X_new = np.random.uniform(0, 1, (1000, 50))
+    y_new = np.random.uniform(0, 1, (1000, 5))
+    
+    cebra_model.fit(Xs, ys)
+    
+    with pytest.raises(NotImplementedError, match=".*multisession.*"):
+        cebra_model.fit(X_new, y_new, adapt=True)
+    
+
 @_util.parametrize_slow(
     arg_names="model_architecture,device,pad_before_transform",
     fast_arguments=list(