diff --git a/SupervisedGym/SupervisedGym/__main__.py b/SupervisedGym/SupervisedGym/__main__.py
index cd061ce1498351f1605b63ba09d1cf905f14f8e4..e03c5da36bff155fa83eb74d0278943478901845 100644
--- a/SupervisedGym/SupervisedGym/__main__.py
+++ b/SupervisedGym/SupervisedGym/__main__.py
@@ -6,11 +6,11 @@ import torch
 import pandas as pd
 import xml.etree.ElementTree as ET
 from SupervisedGym.model_assembler import ModelAssembler
-from SupervisedGym.models import FeedForwardTunable
+from SupervisedGym.models import FeedForwardTunable, StackedRNNTunable, TransformerTunable
 
 
-SAVE_PATH = "<model save path>"
-TIMEOUT = 60 * 60 * 10  # 10 hours minutes
+# SAVE_PATH = "<model save path>"
+# TIMEOUT = 60 * 60 * 10  # 10 hours minutes
 DATA_CONF = DataConf(
     group_id="experimentId",
     pred_horizon=10,
@@ -18,25 +18,44 @@ DATA_CONF = DataConf(
     shuffle=False,
 )
 
+DIRECTORY = r"C:\Users\wojci\OneDrive\Documents\Studia\ZPP\last1\genom"
+
 if __name__ == "__main__":
+
     assebler = ModelAssembler(
         raw_data=pd.read_csv(
-            "/home/szysad/mimuw/3rok/ZPP/training-data/FCR-data/set5/SD-01-29-2021.csv"
+            # "/home/szysad/mimuw/3rok/ZPP/training-data/FCR-data/set5/SD-01-29-2021.csv"
+            r"C:\Users\wojci\OneDrive\Documents\Studia\ZPP\trening\alpha-star-solver\FCRtraining\Genom\genom-newest-at-05-28.csv"
         ),
         cp_problem=ET.parse(
-            "/home/szysad/mimuw/3rok/ZPP/cp_solver/FCR-config/FCRclear-CP.xmi"
+            # "/home/szysad/mimuw/3rok/ZPP/cp_solver/FCR-config/FCRclear-CP.xmi"
+            r"C:\Users\wojci\OneDrive\Documents\Studia\ZPP\trening\alpha-star-solver\FCRtraining\Genom\GenomZPPSolverSmall-CP.xmi"
         ),
         val_loss=torch.nn.L1Loss(),
         data_conf=DATA_CONF,
     )
 
-    model = assebler.assemble_model(
-        timeout=TIMEOUT,
-        n_trials=5,
-        ensemble_size=5,
-        avaible_model_types=[FeedForwardTunable],
-        fast_dev_run=True,
-    )
+    for time in [60 * 60, 5 * 60 * 60]:
+        for networks, name in [
+            ([FeedForwardTunable], "FF"),
+            ([StackedRNNTunable], "RNN"),
+            ([TransformerTunable], "Trans"),
+            ([FeedForwardTunable, StackedRNNTunable, TransformerTunable], "All")
+            ]:
+            for ensembling in [1, 5]:
+                
+                time_per_model = time // len(networks)
+                models = {}
+                for model_type in networks:
+                    models[model_type] = {"timeout": time_per_model, "n_trials": None}
+
+                model = assebler.assemble_model(
+                    ensemble_size=ensembling,
+                    models=models
+                )
+
+                file_name = "Time=" + str(time) + " Model=" + name + " Ensembling=" + str(ensembling)
+                file_path = DIRECTORY + "\\" + file_name
 
-    with open(SAVE_PATH, "wb") as f:
-        f.write(model.parse_model())
+                with open(file_path, "wb") as f:
+                    f.write(model.parse_model())
diff --git a/SupervisedGym/SupervisedGym/modeltuner/objective.py b/SupervisedGym/SupervisedGym/modeltuner/objective.py
index 22abe26eb70d4f66db5bdb67a5297a35169389d4..bec89ed9d62d1aa226db504c24f0c60cf7e5f586 100644
--- a/SupervisedGym/SupervisedGym/modeltuner/objective.py
+++ b/SupervisedGym/SupervisedGym/modeltuner/objective.py
@@ -101,7 +101,7 @@ class Objective:
 
     def __call__(self, trial: optuna.trial.Trial) -> float:
         # generate batch size pow
-        batch_size_pow = trial.suggest_int("batch_size_pow", 5, 12)
+        batch_size_pow = trial.suggest_int("batch_size_pow", 5, 8)
         batch_size = 2 ** batch_size_pow
 
         input_scaler_name = trial.suggest_categorical(
diff --git a/SupervisedGym/SupervisedGym/modeltuner/utils.py b/SupervisedGym/SupervisedGym/modeltuner/utils.py
index 8b9fe625341a468e9760523be6f1ee58b5314995..ea7666265da7e152e535e41a8a1cc2ea4b7e38f4 100644
--- a/SupervisedGym/SupervisedGym/modeltuner/utils.py
+++ b/SupervisedGym/SupervisedGym/modeltuner/utils.py
@@ -51,7 +51,7 @@ def generate_objective_hparams(
     )
     min_epochs = max(1, int(log(train_items, 100)))
     return ObjectiveHParams(
-        val_check_interval=1,
+        val_check_interval=1.0,
         max_epochs=min_epochs * 50,
         min_epochs=min_epochs,
         es_divergence_threshold=None,