diff --git a/docs/mindformers/docs/source_en/usage/evaluation.md b/docs/mindformers/docs/source_en/usage/evaluation.md
index 31e8924df215c6f29914a06aba421937e434bd28..769551f360217af18f4004ac022ea3621baa1b20 100644
--- a/docs/mindformers/docs/source_en/usage/evaluation.md
+++ b/docs/mindformers/docs/source_en/usage/evaluation.md
@@ -240,6 +240,16 @@ Users can follow the following steps to compile and install:
 
 #### Launching a Single-Card Evaluation Script
 
+Before performing the evaluation, please set the environment variables to exclude the influence of protobuf version on the evaluation. Execute the following command:
+
+```shell
+export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python
+```
+
+Execute script path: [eval_with_vlmevalkit.py](https://gitee.com/mindspore/mindformers/blob/dev/toolkit/benchmarks/eval_with_vlmevalkit.py)
+
+Execute the following command for evaluation:
+
 ```shell
 #!/bin/bash
 
@@ -252,8 +262,6 @@ python eval_with_vlmevalkit.py \
   --config-path /{path}/config_path
 ```
 
-Execute script path: [eval_with_vlmevalkit.py](https://gitee.com/mindspore/mindformers/blob/dev/toolkit/benchmarks/eval_with_vlmevalkit.py)
-
 #### Evaluation Parameters
 
 | Parameters            | Type  | Descriptions                                                                                                                                        | Compulsory(Y/N)|
diff --git a/docs/mindformers/docs/source_zh_cn/usage/evaluation.md b/docs/mindformers/docs/source_zh_cn/usage/evaluation.md
index fe2a068f8ecba5942e6ebb67abc7fbf373b277ba..2cd80debd957d1bb5a201c092fce0d290cc203d6 100644
--- a/docs/mindformers/docs/source_zh_cn/usage/evaluation.md
+++ b/docs/mindformers/docs/source_zh_cn/usage/evaluation.md
@@ -239,6 +239,16 @@ Harness评测支持单机单卡、单机多卡、多机多卡场景，每种场
 
 #### 启动单卡评测脚本
 
+执行评测前请先设置环境变量，排除protobuf版本对评测的影响，执行以下命令：
+
+```shell
+export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python
+```
+
+执行脚本路径：[eval_with_vlmevalkit.py](https://gitee.com/mindspore/mindformers/blob/dev/toolkit/benchmarks/eval_with_vlmevalkit.py)
+
+执行以下命令进行评测：
+
 ```shell
 #!/bin/bash
 
@@ -251,8 +261,6 @@ python eval_with_vlmevalkit.py \
   --config-path /{path}/config_path
 ```
 
-执行脚本路径：[eval_with_vlmevalkit.py](https://gitee.com/mindspore/mindformers/blob/dev/toolkit/benchmarks/eval_with_vlmevalkit.py)
-
 #### 评测参数
 
 | 参数            | 类型  | 参数介绍                             | 是否必须 |