diff --git a/tools/train.py b/tools/train.py
index 64dddb7..387649a 100644
--- a/tools/train.py
+++ b/tools/train.py
@@ -7,9 +7,10 @@ from torch.nn.parallel import DistributedDataParallel as DDP
 
 from typing import List, Optional
 from functools import partial
+from khaosz.data import DatasetLoader
 from khaosz.config import ModelParameter, TrainConfig, CosineScheduleConfig
 from khaosz.trainer import Trainer, SchedulerFactory
-from khaosz.data import DatasetLoader
+from khaosz.parallel import get_rank
 
 
 def parse_args() -> argparse.Namespace:
@@ -58,7 +59,7 @@ def parse_args() -> argparse.Namespace:
     return args
 
 def ddp_wrap(model: nn.Module):
-    local_rank = int(os.environ.get("LOCAL_RANK", 0))
+    local_rank = get_rank()
     model = model.to(device=f"cuda:{local_rank}", dtype=torch.bfloat16)
     ddp_model = DDP(
         model,