diff --git a/autotm/algorithms_for_tuning/genetic_algorithm/ga.py b/autotm/algorithms_for_tuning/genetic_algorithm/ga.py
index 729e521..7c1b43a 100644
--- a/autotm/algorithms_for_tuning/genetic_algorithm/ga.py
+++ b/autotm/algorithms_for_tuning/genetic_algorithm/ga.py
@@ -195,6 +195,8 @@ def _sort_population(population):
         population.sort(key=operator.attrgetter("fitness_value"), reverse=True)
 
     def _calculate_uncertain_res(self, generation, iteration_num: int, proc=0.3):
+        if len(generation) == 0:
+            return []
         X = np.array([individ.dto.params.to_vector() for individ in generation])
         certanty = get_prediction_uncertanty(
             self.surrogate.surrogate, X, self.surrogate.name
diff --git a/autotm/algorithms_for_tuning/genetic_algorithm/surrogate.py b/autotm/algorithms_for_tuning/genetic_algorithm/surrogate.py
index 7154d27..b8dd683 100644
--- a/autotm/algorithms_for_tuning/genetic_algorithm/surrogate.py
+++ b/autotm/algorithms_for_tuning/genetic_algorithm/surrogate.py
@@ -53,6 +53,13 @@ def __init__(self, surrogate_name, **kwargs):
         self.gpr_kernel = None
 
     def create(self):
+        kernel = self.kwargs["gpr_kernel"]
+        del self.kwargs["gpr_kernel"]
+        gpr_alpha = self.kwargs["gpr_alpha"]
+        del self.kwargs["gpr_alpha"]
+        normalize_y = self.kwargs["normalize_y"]
+        del self.kwargs["normalize_y"]
+
         if self.name == "random-forest-regressor":
             self.surrogate = RandomForestRegressor(**self.kwargs)
         elif self.name == "mlp-regressor":
@@ -70,8 +77,6 @@ def create(self):
             )
         elif self.name == "GPR":  # tune ??
             if not self.gpr_kernel:
-                kernel = self.kwargs["gpr_kernel"]
-                del self.kwargs["gpr_kernel"]
                 if kernel == "RBF":
                     self.gpr_kernel = 1.0 * RBF(1.0)
                 elif kernel == "RBFwithConstant":
@@ -85,8 +90,8 @@ def create(self):
                 elif kernel == "RationalQuadratic":
                     self.gpr_kernel = RationalQuadratic(1.0)
                 self.kwargs["kernel"] = self.gpr_kernel
-                self.kwargs["alpha"] = self.kwargs["gpr_alpha"]
-                del self.kwargs["gpr_alpha"]
+                self.kwargs["alpha"] = gpr_alpha
+                self.kwargs["normalize_y"] = normalize_y
             self.surrogate = GaussianProcessRegressor(**self.kwargs)
         elif self.name == "decision-tree-regressor":
             try:
diff --git a/autotm/visualization/dynamic_tracker.py b/autotm/visualization/dynamic_tracker.py
index ceb9dd1..28cf92e 100644
--- a/autotm/visualization/dynamic_tracker.py
+++ b/autotm/visualization/dynamic_tracker.py
@@ -212,7 +212,11 @@ def get_metric_df(self):
                 cur_df = pd.DataFrame(cur_df_dict)
                 cur_df[GENERATION_COL] = gen
                 dfs.append(cur_df)
-            self.crossover_df = pd.concat(dfs)
+            if len(dfs) > 0:
+                self.crossover_df = pd.concat(dfs)
+            else:
+                warnings.warn("No crossover changes have been found to save", RuntimeWarning)
+                self.crossover_df = pd.DataFrame([])
 
     def write_metrics_to_file(self):
         os.makedirs(self.save_path, exist_ok=True)