diff --git a/docs/chapter02_prerequisite/2.3_autograd.md b/docs/chapter02_prerequisite/2.3_autograd.md
index 3bce733b07e563b2c081f04685fda50eb5e994db..f796f58da1b40dc5d9e8a2c51f19a5af28d37df8 100644
--- a/docs/chapter02_prerequisite/2.3_autograd.md
+++ b/docs/chapter02_prerequisite/2.3_autograd.md
@@ -74,7 +74,7 @@ True
 <SumBackward0 object at 0x118f50cc0>
 ```
 
-## 2.3.2 梯度
+## 2.3.3 梯度
 因为`out`是一个标量，所以调用`backward()`时不需要指定求导变量：
 ``` python
 out.backward() # 等价于 out.backward(torch.tensor(1.))