update README: reorder FlagOS and model download steps

README.md

@@ -54,7 +54,6 @@ FlagEval (Libra)** is a comprehensive evaluation system and open platform for la
 |CII-Bench|66.27|65.75|
 |Blink|62.70|62.37|
 
-
 # User Guide
 
 **Environment Setup**
@@ -68,18 +67,18 @@ FlagEval (Libra)** is a comprehensive evaluation system and open platform for la
 
 ## Operation Steps
 
-### Download Open-source Model Weights
+### Download FlagOS Image
 
 ```bash
-pip install modelscope
-modelscope download --model stepfun-ai/step3 --local_dir /share/step3
-
+docker pull harbor.baai.ac.cn/flagrelease-public/flagrelease_nvidia_step3
 ```
 
-### Download FlagOS Image
+### Download Open-source Model Weights
 
 ```bash
-docker pull harbor.baai.ac.cn/flagrelease-public/flagrelease_nvidia_step3
+pip install modelscope
+modelscope download --model stepfun-ai/step3 --local_dir /share/step3
+
 ```
 
 ### Start the inference service
@@ -96,7 +95,6 @@ flagscale serve step3
 
 ```
 
-
 ## Service Invocation
 
 ### API-based Invocation Script
@@ -151,7 +149,6 @@ We warmly welcome global developers to join us:
 3. Improve technical documentation
 4. Expand hardware adaptation support
 
-
 # License
 
 本模型的权重来源于stepfun-ai/step3，以apache2.0协议https://www.apache.org/licenses/LICENSE-2.0.txt开源。