From 642b566126131f266cb15d75b29aa452949ff3e0 Mon Sep 17 00:00:00 2001
From: WncFht <2130212584@qq.com>
Date: Sat, 28 Dec 2024 16:17:58 +0800
Subject: [PATCH] update index

---
 docs/Blogs/index.md        | 3 ++-
 docs/Blogs/posts/ULIP-2.md | 5 +++--
 2 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/docs/Blogs/index.md b/docs/Blogs/index.md
index 93a0af68..df1926fa 100644
--- a/docs/Blogs/index.md
+++ b/docs/Blogs/index.md
@@ -16,5 +16,6 @@ comments: false
 
 - "一些想法":
     - 关于交流，信息，选择，焦虑 : posts/24-12-23
-
+- "论文笔记":
+    - ULIP-2 : posts/ULIP-2
 {{ END_TOC }}
diff --git a/docs/Blogs/posts/ULIP-2.md b/docs/Blogs/posts/ULIP-2.md
index 83164561..49bfae1e 100644
--- a/docs/Blogs/posts/ULIP-2.md
+++ b/docs/Blogs/posts/ULIP-2.md
@@ -45,9 +45,10 @@ mdate: 2024-12-28
 
 ## Our Pipeline
 
+ULIP-2 采用大型多模态模型，从 3D 形状的整体角度为每个 2D 渲染的图像自动生成详细描述。ULIP-2 利用预先对齐和冻结的视觉语言特征空间来实现三元体模态之间的对齐：整体文本、图像和 3D 点云。预训练后，3D 编码器将用于下游任务。
+
 > [!example]-
-> ULIP-2 采用大型多模态模型，从 3D 形状的整体角度为每个 2D 渲染的图像自动生成详细描述。ULIP-2 利用预先对齐和冻结的视觉语言特征空间来实现三元体模态之间的对齐：整体文本、图像和 3D 点云。预训练后，3D 编码器将用于下游任务。
-![image.png](https://raw.githubusercontent.com/WncFht/picture/main/20241228152142810.png)
+> ![image.png](https://raw.githubusercontent.com/WncFht/picture/main/20241228152142810.png)
 
 ### Key insight