diff --git a/docs/chapter5/第五章 动手搭建大模型.md b/docs/chapter5/第五章 动手搭建大模型.md
index c42a156..ddbf7a5 100644
--- a/docs/chapter5/第五章 动手搭建大模型.md	
+++ b/docs/chapter5/第五章 动手搭建大模型.md	
@@ -114,7 +114,7 @@ torch.Size([1, 50, 768])
 在 LLaMA2 模型中，虽然只有 LLaMA2-70B模型使用了分组查询注意力机制（Grouped-Query Attention，GQA），但我们依然选择使用 GQA 来构建我们的 LLaMA Attention 模块，它可以提高模型的效率，并节省一些显存占用。
 
 <div align='center'>
-    <img src="https://raw.githubusercontent.com/datawhalechina/happy-llm/main/docs/images/5-images/Attention.png" alt="alt text" width="100%">
+    <img src="https://raw.githubusercontent.com/datawhalechina/happy-llm/main/docs/images/5-images/llama2-attention" alt="alt text" width="70%">
     <p>图 5.2 LLaMA2 Attention 结构</p>
 </div>
 
diff --git a/docs/images/5-images/Attention.png b/docs/images/5-images/Attention.png
deleted file mode 100644
index 99070f0..0000000
Binary files a/docs/images/5-images/Attention.png and /dev/null differ
diff --git a/docs/images/5-images/llama2-attention.png b/docs/images/5-images/llama2-attention.png
new file mode 100644
index 0000000..0d3bf83
Binary files /dev/null and b/docs/images/5-images/llama2-attention.png differ