diff --git a/discrete_mixed_bo/experiment_utils.py b/discrete_mixed_bo/experiment_utils.py
index 813e5a7..c623d9d 100644
--- a/discrete_mixed_bo/experiment_utils.py
+++ b/discrete_mixed_bo/experiment_utils.py
@@ -716,8 +716,8 @@ def get_problem(name: str, dim: Optional[int] = None, **kwargs) -> DiscreteTestP
 
 
 def generate_discrete_options(
-    base_function: DiscreteTestProblem,
-) -> List[Dict[int, float]]:
+    base_function: DiscreteTestProblem, return_tensor: bool = False,
+) -> Union[List[Dict[int, float]], Tensor]:
     categorical_features = base_function.categorical_features
     discrete_indices = torch.cat(
         [base_function.integer_indices, base_function.categorical_indices], dim=0
@@ -764,4 +764,6 @@ def generate_discrete_options(
             start_idx = end_idx
     # create a list of dictionaries of mapping indices to values
     # the list has a dictionary for each discrete configuration
+    if return_tensor:
+        return discrete_options.to(base_function.bounds)
     return [dict(zip(indices, xi)) for xi in discrete_options.tolist()]
diff --git a/discrete_mixed_bo/run_one_replication.py b/discrete_mixed_bo/run_one_replication.py
index 689aae9..2549bea 100644
--- a/discrete_mixed_bo/run_one_replication.py
+++ b/discrete_mixed_bo/run_one_replication.py
@@ -67,7 +67,7 @@
     "exact_round__fin_diff__nehvi-1",
     "exact_round__ste__nehvi-1",
     "enumerate__nehvi-1",
-    "nevergrad_porfolio",
+    "nevergrad_portfolio",
 ]
 
 
@@ -405,7 +405,7 @@ def run_one_replication(
                 # construct a list of dictionaries mapping indices in one-hot space
                 # to parameter values.
                 discrete_options = generate_discrete_options(
-                    base_function=base_function,
+                    base_function=base_function, return_tensor=base_function.cont_indices.shape[0]==0,
                 )
                 if base_function.cont_indices.shape[0] > 0:
                     # optimize mixed