From 2fe5e0b4f8bdc5df4fc0def5939d8838618ffe1e Mon Sep 17 00:00:00 2001
From: Zhengkai Jiang <kaikaijiang.jzk@gmail.com>
Date: Wed, 28 Feb 2024 23:18:26 +0800
Subject: [PATCH] Update README.md

---
 Large-DiT/README.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/Large-DiT/README.md b/Large-DiT/README.md
index 8bf2834..5b4bcc8 100644
--- a/Large-DiT/README.md
+++ b/Large-DiT/README.md
@@ -3,7 +3,7 @@
 We release the ***Large Diffusion Transformer*** (**L-DiT-3B** & **L-DiT-7B** 🔥), inspired by the methodologies of [LLaMA](https://github.com/facebookresearch/llama) and [DiT](https://github.com/facebookresearch/DiT). 
 
 ![image](./assets/sample_t2i.png)
-![image](./assets/sample2.jpg)
+![image](./assets/sample.png)
 
 
 Compared to DiT-XL/2, our L-DiT-7B:
@@ -26,7 +26,7 @@ We observe instability issues during the training of the original DiT, particula
 ### ImageNet 256x256 Samples and Benchmark
 
 <div>
-  <img src="./assets/sample.png"/ width="57%">  <img src="./assets/table.png"/ width="37%"> <br>
+  <img src="./assets/sample.png"/ width="62%">  <img src="./assets/table.png"/ width="32%"> <br>
 </div>