diff --git a/README.md b/README.md
index 0b4815239b1..d18d56420b2 100644
--- a/README.md
+++ b/README.md
@@ -1,7 +1,7 @@
 
 ## Quick Start
 
-### Interactive Mode
+### Basic Setup
 
 1. Install Python and packages
 
@@ -10,31 +10,39 @@ Note: Python 3.5+ is required.
 
 Install packages dicttoxml gitpython lxml networkx numpy openpyxl statistics scipy matplotlib: <https://packaging.python.org/installing/>
 
-E.g., on Ubuntu:
+E.g., on Ubuntu LTS 16.04:
 ```
+sudo apt update
 sudo apt install -y python3 python3-pip
 sudo -H pip3 install dicttoxml gitpython lxml networkx numpy openpyxl statistics scipy matplotlib
 ```
 
+2. Update Git
+
 In order to uset the `--indent-heuristic` option of `git diff`, we require git version >= 2.11. Use the following commands to upgrade:
 ```
+git --version
 sudo add-apt-repository ppa:git-core/ppa -y
 sudo apt-get update
 sudo apt-get install git -y
 git --version
 ```
 
+3. Apply a patch to gitpython
+
 (Try to) apply a patch to gitpython 2.1.x:
 ```
 cd misc/
 ./apply_patch.py
 ```
 
-2. Install Jupyter
+### Interactive Mode
+
+1. Install Jupyter
 
 Reference <http://jupyter.org/install.html>.
 
-E.g., on Ubuntu:
+E.g., on Ubuntu LTS 16.04:
 ```
 sudo -H pip3 install --upgrade pip
 sudo -H pip3 install jupyter
@@ -46,21 +54,7 @@ sudo apt install -y jq
 ./gitconfig.sh
 ```
 
-3. Install TensorFlow
-
-Follow this tutorial: https://www.tensorflow.org/install/install_sources.
-
-An example build:
-
-```
-bazel build --config=opt --copt=-mavx --copt=-mavx2 --copt=-mfma --copt=-msse4.1 --copt=-msse4.2 //tensorflow/tools/pip_package:build_pip_package
-```
-
-When running TensorFlow, get out of the tensorflow source dir. Otherwise,
-python would prompt an error message "No module named
-pywrap_tensorflow_internal".
-
-4. Run a notebook
+2. Run a notebook
 
 All notebooks should be run in the root dir of this project.
 
@@ -73,7 +67,7 @@ Enjoy your interactions with the notebook!
 
 ### Batch Mode
 
-Do the above Step 1 and Step 2.
+Complete the basic setup first.
 Note: setup-linux-ubuntu.sh can be used for Ubuntu Server.
 
 Read help info of dev_analysis.py:
@@ -95,3 +89,18 @@ A sample long-time run:
 ```
 nohup ./dev_analysis.py -s ./repos/linux/ -x ./repos/linux-4.10-xml/ -o linux-4.10-cc.xlsx -n 1000 10000 -a 0 1 0.05 -c > dev.out 2>&1 &
 ```
+
+### Install TensorFlow
+
+Follow this tutorial: https://www.tensorflow.org/install/install_sources.
+
+An example build:
+
+```
+bazel build --config=opt --copt=-mavx --copt=-mavx2 --copt=-mfma --copt=-msse4.1 --copt=-msse4.2 //tensorflow/tools/pip_package:build_pip_package
+```
+
+When running TensorFlow, get out of the tensorflow source dir. Otherwise,
+python would prompt an error message "No module named
+pywrap_tensorflow_internal".
+