[App] Resolve multi-node cloud bug (#15619)

update

(cherry picked from commit 47314ea)

examples/app_multi_node/train_lite.py

@@ -9,7 +9,7 @@ class LitePyTorchDistributed(L.LightningWork):
     @staticmethod
     def run():
         # 1. Create LightningLite.
-        lite = LightningLite(strategy="ddp", precision="bf16")
+        lite = LightningLite(strategy="ddp", precision=16)
 
         # 2. Prepare distributed model and optimizer.
         model = torch.nn.Linear(32, 2)