diff --git a/docs/source/migration.md b/docs/source/migration.md
index d04b66d5..f50d1dff 100644
--- a/docs/source/migration.md
+++ b/docs/source/migration.md
@@ -104,6 +104,6 @@ for batch in train_data:
     loss = model(batch)
     loss.backward()
     torch.nn.utils.clip_grad_norm_(model.parameters(), max_grad_norm)  # Gradient clipping is not in AdamW anymore (so you can use amp without issue)
-    scheduler.step()
     optimizer.step()
+    scheduler.step()
 ```