diff --git a/ernie-doc/README.md b/ernie-doc/README.md
index c9beb1c5939bdf2e193ccb687644798fd2e35f44..dc9b1840cb560034e022bf29f711b6b9a5f39a0b 100644
--- a/ernie-doc/README.md
+++ b/ernie-doc/README.md
@@ -179,7 +179,7 @@ We compare the performance of [ERNIE-Doc](https://arxiv.org/abs/2012.15688) with
 
 ### Install PaddlePaddle
 
-This code base has been tested with Paddle (version>=1.8) with Python3. Other dependency of ERNIE-GEN is listed in `requirements.txt`, you can install it by
+This code base has been tested with Paddle (version>=2.0) with Python3. Other dependency of ERNIE-Doc is listed in `requirements.txt`, you can install it by
 ```script
 pip install -r requirements.txt
 ```
@@ -191,7 +191,7 @@ sh script/run_imdb.sh
 sh script/run_iflytek.sh
 sh script/run_dureader.sh
 ```
-[Preprocessing code for IMDB dataset](./ernie_doc/data/imdb/README.md)
+[Preprocessing code for IMDB dataset](./data/imdb/README.md)
 
 
 The log of training and the evaluation results are in `log/job.log.0`.
diff --git a/ernie-doc/README_zh.md b/ernie-doc/README_zh.md
index 96a151e3a821e9f74ebad63d9574980fbf665856..05dd1700ed6eafd222f908dedba83d38a9413ea5 100644
--- a/ernie-doc/README_zh.md
+++ b/ernie-doc/README_zh.md
@@ -178,7 +178,7 @@
 
 ### 安装飞桨
 
-我们的代码基于 Paddle(version>=1.8)，推荐使用python3运行。 ERNIE-Doc 依赖的其他模块也列举在 `requirements.txt`，可以通过下面的指令安装:
+我们的代码基于 Paddle(version>=2.0)，推荐使用python3运行。 ERNIE-Doc 依赖的其他模块也列举在 `requirements.txt`，可以通过下面的指令安装:
 ```script
 pip install -r requirements.txt
 ```
@@ -190,7 +190,7 @@ sh script/run_imdb.sh # 英文分类任务
 sh script/run_iflytek.sh # 中文分类任务
 sh script/run_dureader.sh # 中文阅读理解任务
 ```
-[imdb数据处理说明](./ernie_doc/data/imdb/README.md)
+[imdb数据处理说明](./data/imdb/README.md)
 
 
 具体微调参数均可在上述脚本中进行修改，训练和评估的日志在 `log/job.log.0`。 
diff --git a/ernie-doc/data/imdb/README.md b/ernie-doc/data/imdb/README.md
index 1c24be38bc4d7b8ecdbfa83fe807ea7e322aefc0..b58aff9f0e16858ccdc1efbba38e337c6bc3a197 100644
--- a/ernie-doc/data/imdb/README.md
+++ b/ernie-doc/data/imdb/README.md
@@ -1,10 +1,10 @@
-## 下载官方数据
+## 下载官方数据 (Download data)
 
 http://ai.stanford.edu/~amaas/data/sentiment/index.html
 
-## 运行预处理脚本
+## 运行预处理脚本 (Run preprocessing code)
 
 ```python
-python multi_files_to_one.py
+python multi_files_to_one.py # this will generate train/test txt
 ```
-生成train.txt与test.txt文件至该文件夹下
\ No newline at end of file
+生成train.txt与test.txt文件至该文件夹下