diff --git a/scripts/run_simpo.py b/scripts/run_simpo.py
index 77374a4..d5fabd2 100644
--- a/scripts/run_simpo.py
+++ b/scripts/run_simpo.py
@@ -26,8 +26,6 @@ from alignment import (
     DPOConfig,
     H4ArgumentParser,
     ModelArguments,
-    maybe_insert_system_message,
-    is_openai_format,
     get_checkpoint,
     get_datasets,
     get_kbit_device_map,
@@ -36,6 +34,7 @@ from alignment import (
     get_tokenizer,
     is_adapter_model,
 )
+from alignment.data import maybe_insert_system_message, is_openai_format
 from peft import PeftConfig, PeftModel
 from simpo_trainer import SimPOTrainer
 from dataclasses import dataclass, field
@@ -167,7 +166,7 @@ def main():
         splits=data_args.dataset_splits,
         configs=data_args.dataset_configs,
         columns_to_keep=["messages", "chosen", "rejected", "prompt", "completion", "label"],
-        seed=training_args.seed,
+        # seed=training_args.seed,
     )
     logger.info(
         f"Training on the following splits: {[split + ' : ' + str(dset.num_rows) for split, dset in raw_datasets.items()]}"