diff --git a/builder/manywheel/entrypoint_build.sh b/builder/manywheel/entrypoint_build.sh
index abb90562a2..8d1eb16de9 100755
--- a/builder/manywheel/entrypoint_build.sh
+++ b/builder/manywheel/entrypoint_build.sh
@@ -11,7 +11,7 @@ source /opt/conda/bin/activate
 conda activate $PYTHON_VERSION
 
 cd lmdeploy
-mkdir build && cd build
+mkdir -p build && cd build && rm -rf *
 bash ../generate.sh
 make -j$(nproc) && make install
 if [ $? != 0 ]; then
diff --git a/docs/en/build.md b/docs/en/build.md
index d8bd85fe85..cb278073c9 100644
--- a/docs/en/build.md
+++ b/docs/en/build.md
@@ -1,27 +1,79 @@
-## Build from source
+# Build from source
 
-- make sure local gcc version no less than 9, which can be conformed by `gcc --version`.
-- install packages for compiling and running:
+LMDeploy provides prebuilt package that can be easily installed by `pip install lmdeploy`.
+
+If you have requests to build lmdeploy from source, please clone lmdeploy repository from GitHub, and follow instructions in next sections
+
+```shell
+git clone --depth=1 https://github.com/InternLM/lmdeploy
+```
+
+## Build in Docker (recommended)
+
+We highly advise using the provided docker image for lmdeploy build to circumvent complex environment setup.
+
+The docker image is `openmmlab/lmdeploy-builder:cuda11.8`. Make sure that docker is installed before using this image.
+
+In the root directory of the lmdeploy source code, please run the following command:
+
+```shell
+cd lmdeploy # the home folder of lmdeploy source code
+bash builder/manywheel/build_all_wheel.sh
+```
+
+All the wheel files for lmdeploy under py3.8 - py3.11 will be found in the `builder/manywheel/cuda11.8_dist` directory, such as,
+
+```text
+builder/manywheel/cuda11.8_dist/
+├── lmdeploy-0.0.12-cp310-cp310-manylinux2014_x86_64.whl
+├── lmdeploy-0.0.12-cp311-cp311-manylinux2014_x86_64.whl
+├── lmdeploy-0.0.12-cp38-cp38-manylinux2014_x86_64.whl
+└── lmdeploy-0.0.12-cp39-cp39-manylinux2014_x86_64.whl
+```
+
+If the wheel file for a specific Python version is required, such as py3.8, please execute:
+
+```shell
+bash builder/manywheel/build_wheel.sh py38 manylinux2014_x86_64 cuda11.8 cuda11.8_dist
+```
+
+And the wheel file will be found in the `builder/manywheel/cuda11.8_dist` directory.
+
+You can use `pip install` to install the wheel file that matches the Python version on your host machine.
+
+## Build in localhost (optional)
+
+Firstly, please make sure gcc version is no less than 9, which can be conformed by `gcc --version`.
+
+Then, follow the steps below to set up the compilation environment:
+
+- install the dependent packages:
   ```shell
   pip install -r requirements.txt
+  apt-get install rapidjson-dev
   ```
-- install [nccl](https://docs.nvidia.com/deeplearning/nccl/install-guide/index.html), set environment variables:
+- install [nccl](https://docs.nvidia.com/deeplearning/nccl/install-guide/index.html), and set environment variables:
   ```shell
   export NCCL_ROOT_DIR=/path/to/nccl/build
   export NCCL_LIBRARIES=/path/to/nccl/build/lib
   ```
-- install rapidjson
-- install openmpi, installing from source is recommended.
+- install openmpi from source:
   ```shell
   wget https://download.open-mpi.org/release/open-mpi/v4.1/openmpi-4.1.5.tar.gz
-  tar -xzf openmpi-*.tar.gz && cd openmpi-*
-  ./configure --with-cuda
-  make -j$(nproc)
-  make install
+  tar xf openmpi-4.1.5.tar.gz
+  cd openmpi-4.1.5
+  ./configure
+  make -j$(nproc) && make install
   ```
-- build and install lmdeploy:
+- build and install lmdeploy libraries:
   ```shell
+  cd lmdeploy # the home folder of lmdeploy
   mkdir build && cd build
   sh ../generate.sh
   make -j$(nproc) && make install
   ```
+- install lmdeploy python package:
+  ```shell
+  cd ..
+  pip install -e .
+  ```
diff --git a/docs/zh_cn/build.md b/docs/zh_cn/build.md
index 6763adec47..a73296354b 100644
--- a/docs/zh_cn/build.md
+++ b/docs/zh_cn/build.md
@@ -1,27 +1,79 @@
-### 源码安装
+# 编译和安装
+
+LMDeploy 提供了预编译包，可以很方便的通过 `pip install lmdeploy` 安装和使用。
+
+如果有源码编译的需求，请先下载 lmdeploy 源码：
+
+```shell
+git clone --depth=1 https://github.com/InternLM/lmdeploy
+```
+
+然后，参考以下章节编译和安装。
+
+## 在 docker 内编译安装（强烈推荐）
+
+LMDeploy 提供了编译镜像 `openmmlab/lmdeploy-builder:cuda11.8`。使用之前，请确保 docker 已安装。
+
+在 lmdeploy 源码的根目录下，运行以下命令：
+
+```shell
+cd lmdeploy # lmdeploy 源码根目录
+bash builder/manywheel/build_all_wheel.sh
+```
+
+即可在 `builder/manywheel/cuda11.8_dist` 文件夹下，得到 lmdeploy 在 py3.8 - py3.11 下所有的 wheel 文件。比如，
+
+```text
+builder/manywheel/cuda11.8_dist/
+├── lmdeploy-0.0.12-cp310-cp310-manylinux2014_x86_64.whl
+├── lmdeploy-0.0.12-cp311-cp311-manylinux2014_x86_64.whl
+├── lmdeploy-0.0.12-cp38-cp38-manylinux2014_x86_64.whl
+└── lmdeploy-0.0.12-cp39-cp39-manylinux2014_x86_64.whl
+```
+
+如果需要固定 python 版本的 wheel 文件，比如 py3.8，可以执行：
+
+```shell
+bash builder/manywheel/build_wheel.sh py38 manylinux2014_x86_64 cuda11.8 cuda11.8_dist
+```
+
+wheel 文件存放在目录 `builder/manywheel/cuda11.8_dist` 下。
+
+在宿主机上，通过 `pip install` 安装和宿主机python版本一致的 wheel 文件，即完成 lmdeploy 整个编译安装过程。
+
+## 在物理机上编译安装（可选）
+
+首先，请确保物理机环境的 gcc 版本不低于 9，可以通过`gcc --version`确认。
+
+然后，按如下步骤，配置编译环境：
 
-- 确保物理机环境的 gcc 版本不低于 9，可以通过`gcc --version`确认。
 - 安装编译和运行依赖包：
   ```shell
   pip install -r requirements.txt
+  apt-get install rapidjson-dev
   ```
 - 安装 [nccl](https://docs.nvidia.com/deeplearning/nccl/install-guide/index.html),设置环境变量
   ```shell
   export NCCL_ROOT_DIR=/path/to/nccl/build
   export NCCL_LIBRARIES=/path/to/nccl/build/lib
   ```
-- rapidjson 安装
-- openmpi 安装, 推荐从源码安装:
+- 源码编译安装 openmpi:
   ```shell
   wget https://download.open-mpi.org/release/open-mpi/v4.1/openmpi-4.1.5.tar.gz
-  tar -xzf openmpi-*.tar.gz && cd openmpi-*
-  ./configure --with-cuda
-  make -j$(nproc)
-  make install
+  tar xf openmpi-4.1.5.tar.gz
+  cd openmpi-4.1.5
+  ./configure
+  make -j$(nproc) && make install
   ```
 - lmdeploy 编译安装:
   ```shell
+  cd lmdeploy # lmdeploy 源码的根目录
   mkdir build && cd build
   sh ../generate.sh
   make -j$(nproc) && make install
   ```
+- 安装 lmdeploy python package:
+  ```shell
+  cd ..
+  pip install -e .
+  ```