diff --git a/train.py b/train.py
index 7cfee808..f1a955ee 100644
--- a/train.py
+++ b/train.py
@@ -146,7 +146,7 @@ def main(args):
     # Note that parameter initialization is done within the DiT constructor
     ema = deepcopy(model).to(device)  # Create an EMA of the model for use after training
     requires_grad(ema, False)
-    model = DDP(model.to(device), device_ids=[rank])
+    model = DDP(model.to(device))
     diffusion = create_diffusion(timestep_respacing="")  # default: 1000 steps, linear noise schedule
     vae = AutoencoderKL.from_pretrained(f"stabilityai/sd-vae-ft-{args.vae}").to(device)
     logger.info(f"DiT Parameters: {sum(p.numel() for p in model.parameters()):,}")