diff --git a/src/gurobi_ml/modeling/decision_tree/decision_tree_model.py b/src/gurobi_ml/modeling/decision_tree/decision_tree_model.py
index c8f1e7cb..62b2e182 100644
--- a/src/gurobi_ml/modeling/decision_tree/decision_tree_model.py
+++ b/src/gurobi_ml/modeling/decision_tree/decision_tree_model.py
@@ -126,6 +126,8 @@ def _leaf_formulation(
     # We should attain 1 leaf
     gp_model.addConstr(leafs_vars.sum(axis=1) == 1)
 
+    gp_model.addConstr(output <= np.max(tree["value"], axis=0))
+    gp_model.addConstr(output >= np.min(tree["value"], axis=0))
     if verbose:
         timer.timing(f"Added {nex} linear constraints")
 
@@ -207,8 +209,8 @@ def _paths_formulation(gp_model, _input, output, tree, epsilon, _name_var):
             for i in range(outdim)
         )
 
-    output.LB = np.min(tree.value)
-    output.UB = np.max(tree.value)
+    gp_model.addConstr(output <= np.max(tree["value"], axis=0))
+    gp_model.addConstr(output >= np.min(tree["value"], axis=0))
 
 
 class AbstractTreeEstimator(AbstractPredictorConstr):
diff --git a/tests/test_lightgbm/test_lightgbm_formulations.py b/tests/test_lightgbm/test_lightgbm_formulations.py
index 8d264712..7eda3afe 100644
--- a/tests/test_lightgbm/test_lightgbm_formulations.py
+++ b/tests/test_lightgbm/test_lightgbm_formulations.py
@@ -14,7 +14,7 @@ class TestLGBMhModel(FixedRegressionModel):
 
     basedir = os.path.join(os.path.dirname(__file__), "..", "predictors")
 
-    def test_diabetes_xgboost_pairs(self):
+    def test_diabetes_lightgbm_pairs(self):
         data = datasets.load_diabetes()
         X = data["data"]
         y = data["target"]
@@ -25,7 +25,7 @@ def test_diabetes_xgboost_pairs(self):
 
         self.do_one_case(one_case, X, 6, "pairs", float_type=np.float32)
 
-    def test_diabetes_xgboost_pairs_pipeline(self):
+    def test_diabetes_lightgbm_pairs_pipeline(self):
         data = datasets.load_diabetes()
         X = data["data"]
         y = data["target"]
@@ -37,7 +37,7 @@ def test_diabetes_xgboost_pairs_pipeline(self):
 
         self.do_one_case(one_case, X, 6, "pairs", float_type=np.float32)
 
-    def test_diabetes_xgboost_all(self):
+    def test_diabetes_lightgbm_all(self):
         data = datasets.load_diabetes()
         X = data["data"]
         y = data["target"]