From 774e5073c125eb29ce003aab2e3fb9732efa893d Mon Sep 17 00:00:00 2001
From: Logan Zou <74288839+logan-zou@users.noreply.github.com>
Date: Wed, 28 May 2025 10:39:57 +0800
Subject: [PATCH] =?UTF-8?q?Update=20=E7=AC=AC=E4=BA=8C=E7=AB=A0=20Transfor?=
 =?UTF-8?q?mer=E6=9E=B6=E6=9E=84.md?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 docs/chapter2/第二章 Transformer架构.md | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/docs/chapter2/第二章 Transformer架构.md b/docs/chapter2/第二章 Transformer架构.md
index a3c54d0..bbcd296 100644
--- a/docs/chapter2/第二章 Transformer架构.md	
+++ b/docs/chapter2/第二章 Transformer架构.md	
@@ -9,10 +9,12 @@
 - 前馈神经网络（Feedforward Neural Network，FNN），即每一层的神经元都和上下两层的每一个神经元完全连接，如图2.1所示:
 
 <div align="center">
-  <img src="https://github.com/datawhalechina/happy-llm/blob/main/docs/images/2-figures/1-0.png" alt="图片描述" width="90%"/>
+  <img src="../images/2-figures/1-0.png" alt="图片描述" width="90%"/>
   <p>图2.1 前馈神经网络</p>
 </div>
 
+![](../images/2-figures/1-0.png)
+
 - 卷积神经网络（Convolutional Neural Network，CNN），即训练参数量远小于前馈神经网络的卷积层来进行特征提取和学习，如图2.2所示:
 
 <div align="center">