diff --git a/causy/graph_model.py b/causy/graph_model.py
index e7a6228..a98018a 100644
--- a/causy/graph_model.py
+++ b/causy/graph_model.py
@@ -34,7 +34,6 @@ class AbstractGraphModel(GraphModelInterface, ABC):
     pipeline_steps: List[PipelineStepInterface]
     graph: BaseGraphInterface
     pool: mp.Pool
-    initialize_pool_fn: Callable = lambda self: mp.Pool(mp.cpu_count() * 2)
 
     def __init__(
         self,
@@ -44,10 +43,24 @@ def __init__(
         self.graph = graph
         self.pipeline_steps = pipeline_steps or []
         if self.__multiprocessing_required(self.pipeline_steps):
-            self.pool = self.initialize_pool_fn()
+            self.pool = self.__initialize_pool()
         else:
             self.pool = None
 
+    def __initialize_pool(self) -> mp.Pool:
+        """
+        Initialize the multiprocessing pool
+        :return: the multiprocessing pool
+        """
+        # we need to set the start method to spawn because the default fork method does not work well with torch
+        try:
+            mp.set_start_method("spawn")
+        except RuntimeError:
+            logger.warning(
+                "Could not set multiprocessing start method to spawn. Using default method."
+            )
+        return mp.Pool(mp.cpu_count() * 2)
+
     def __multiprocessing_required(self, pipeline_steps):
         """
         Check if multiprocessing is required
@@ -244,7 +257,7 @@ def execute_pipeline_step(self, test_fn: PipelineStepInterface):
                 logger.warning(
                     "Parallel processing is enabled but no pool is initialized. Initializing pool."
                 )
-                self.pool = self.initialize_pool_fn()
+                self.pool = self.__initialize_pool()
 
             for result in self.pool.imap_unordered(
                 unpack_run,