Do not pass custom_tasks by default

src/backend/run_eval_suite_lighteval.py

@@ -66,7 +66,7 @@ def run_evaluation(
         max_samples=limit,
         use_chat_template=False,
         system_prompt=None,
-        custom_tasks_directory="custom_tasks.py",  # if using a custom task
+        # custom_tasks_directory="custom_tasks.py",  # TODO: pass if using a custom task
     )
 
     model_config = InferenceEndpointModelConfig(