diff --git a/demo/demo_cn.rst b/demo/demo_cn.rst
new file mode 100644
index 0000000000000000000000000000000000000000..fa8a1924aee003d3eb206db5654faa64d42a5359
--- /dev/null
+++ b/demo/demo_cn.rst
@@ -0,0 +1,12 @@
+示例
+=====
+
+VisualDL 支持 Python 和 C++ 基于 DL 框架,
+下面是不同平台的一些示例。
+
+.. toctree::
+ :maxdepth: 1
+
+ keras/TUTORIAL_CN.md
+ mxnet/TUTORIAL_CN.md
+ pytorch/TUTORIAL_CN.md
\ No newline at end of file
diff --git a/demo/demo_en.rst b/demo/demo_en.rst
new file mode 100644
index 0000000000000000000000000000000000000000..36cd75f1ba2ca0c0f714535235315dbdb7c773a0
--- /dev/null
+++ b/demo/demo_en.rst
@@ -0,0 +1,12 @@
+Examples
+========
+
+VisualDL supports Python and C++ based DL frameworks,
+here are some examples for different platforms.
+
+.. toctree::
+ :maxdepth: 1
+
+ keras/TUTORIAL_EN.md
+ mxnet/TUTORIAL_EN.md
+ pytorch/TUTORIAL_EN.md
\ No newline at end of file
diff --git a/demo/keras/TUTORIAL_CN.md b/demo/keras/TUTORIAL_CN.md
index 7f93fbd4e2d4b86e45007cea7e88350ebfef296b..2e6bc0e4a066a1c59a7380e9c8343b0b84803d03 100644
--- a/demo/keras/TUTORIAL_CN.md
+++ b/demo/keras/TUTORIAL_CN.md
@@ -92,4 +92,4 @@ class LossHistory(keras.callbacks.Callback):
-完整的演示程序可以在[这里](./keras_mnist_demo.py)下载。
+完整的演示程序可以在[这里](https://github.com/PaddlePaddle/VisualDL/blob/develop/demo/keras/keras_mnist_demo.py)下载。
diff --git a/demo/keras/TUTORIAL_EN.md b/demo/keras/TUTORIAL_EN.md
index 79c8bf60850c3c4efadb7bae2b99d5bc2396f6e0..74a1cd31d8d543021c24113893337e59dc86e3fd 100644
--- a/demo/keras/TUTORIAL_EN.md
+++ b/demo/keras/TUTORIAL_EN.md
@@ -87,4 +87,4 @@ The histograms of the training parameters is as follows:
-The full demonstration code can be downloaded in [here](./keras_mnist_demo.py).
\ No newline at end of file
+The full demonstration code can be downloaded in [here](https://github.com/PaddlePaddle/VisualDL/blob/develop/demo/keras/keras_mnist_demo.py).
\ No newline at end of file
diff --git a/demo/mxnet/TUTORIAL_CN.md b/demo/mxnet/TUTORIAL_CN.md
index b108b678487c0d43a4aba6aaeacdcbe7aa5efa29..d9cec10fda4b9da9e9fa5512199298bd709bc8c5 100644
--- a/demo/mxnet/TUTORIAL_CN.md
+++ b/demo/mxnet/TUTORIAL_CN.md
@@ -59,7 +59,7 @@ lenet_model.fit(train_iter,
这样就好了。在MXNet的训练过程中,每一个批次(batch)训练完后,都会调用我们的回调函数来对准确率进行记录。如您所料,随着训练的进行,准确率会不断上升直到95%以上。以下是两个epoch训练过后的准确率走向:
-![](./epoch2_small.png)
+![](https://github.com/PaddlePaddle/VisualDL/blob/develop/demo/mxnet/epoch2_small.png?raw=true)
## 用VisualDL展示模型图
diff --git a/demo/mxnet/TUTORIAL_EN.md b/demo/mxnet/TUTORIAL_EN.md
index 64e38369946011b3ae55af374487943c1e2738c8..a002579859711a46e373b6587b6a6bacf3287d54 100644
--- a/demo/mxnet/TUTORIAL_EN.md
+++ b/demo/mxnet/TUTORIAL_EN.md
@@ -61,7 +61,7 @@ That's all. In the training process of MXNet, our callback function is called to
The rate of accuracy will continue to rise until more than 95%.
The following is the accuracy of the two epochs:
-![](./epoch2_small.png)
+![](https://github.com/PaddlePaddle/VisualDL/blob/develop/demo/mxnet/epoch2_small.png?raw=true)
## Display the model graph with VisualDL
diff --git a/docs/demo b/docs/demo
new file mode 120000
index 0000000000000000000000000000000000000000..337bec7b50c327580adad4bcbcf8c92d602bc9fa
--- /dev/null
+++ b/docs/demo
@@ -0,0 +1 @@
+../demo/
\ No newline at end of file
diff --git a/docs/index_cn.rst b/docs/index_cn.rst
index 245de6f30960f8df30bbd7427e43edbe2a608de4..9c135217dbaccef6f7a7bf79381f79798f8eaf0b 100644
--- a/docs/index_cn.rst
+++ b/docs/index_cn.rst
@@ -7,4 +7,5 @@ VisualDL
首页
quick_start_cn.md
contribute_code_cn.md
+ demo/demo_cn.rst
api/visualdl.rst
\ No newline at end of file
diff --git a/docs/index_en.rst b/docs/index_en.rst
index 9bec6c4c01f7806dd9f42d2c8a451f3e733a0faa..307b3128ab684b19017eb32dc684aec58d2e7d12 100644
--- a/docs/index_en.rst
+++ b/docs/index_en.rst
@@ -8,4 +8,5 @@ VisualDL
quick_start_en.md
contribute_code_en.md
how_to_dev_frontend_en.md
+ demo/demo_en.rst
api/visualdl.rst
\ No newline at end of file