readme minor update (#5)

Signed-off-by: roger feng <[email protected]>

Author: rogerxfeng8

README.md

@@ -1,46 +1,45 @@
 ## About
 
-This repo is designed as a vLLM plugin for Intel GPU (also know as XPU in pytorch) which will provide custom kernels used in vLLM.
+This repo is designed as a vLLM plugin which provides custom kernels for Intel GPU (known as XPU in PyTorch).
 
 ## Getting started
+Currently we use PyTorch 2.8, oneapi 2025.1.
 
-current we use torch 2.8, oneapi is 2025.1.
+### How it works
+python3 setup.py build - will build a `_C.abi3.so` under build directory
+python3 setup.py install - will copy above .so to `vllm_xpu_kernels` folder
+python3 setup.py develop - will be local install if we use develop build or system/virtual env lib path if we use install.
 
-### how this project works
-Build system of this project will build a `_C.abi3.so` under build dir, and install step will copy this to `vllm_xpu_kernels` folder(will be local install if we use develop build or system/virtual env lib path if we use install). On vllm side, we will
-`import vllm_xpu_kernels._C` at start time which should register all custom ops so we can directly use.
+On vllm side, we will `import vllm_xpu_kernels._C` at start time which should register all custom ops so we can directly use.
 
 ### prepare
 
 Install oneapi 2025.1 deep learning essential [dependency](https://www.intel.com/content/www/us/en/developer/tools/oneapi/base-toolkit-download.html).
 
-create a new virtual env, install build dependency and torch dependency
+Create a new virtual env, install build dependency and torch dependency
 
 ```
 pip install -r requirements.txt
 ```
 
 ### build & install
-development install command (which will install to current dir):
+Build development installation to current directory:
 
 ```
 VLLM_TARGET_DEVICE=xpu python3 setup.py develop
 ```
 
-or
-
-install to system dir
+or installation to system directory:
 
 ```
 VLLM_TARGET_DEVICE=xpu python3 setup.py install
 ```
 
-or build wheel(generate on dist folder)
+or build wheel (generated .whl in dist folder)
 
 ```
 VLLM_TARGET_DEVICE=xpu python3 setup.py bdist_wheel
 ```
 
 ### how to use in vLLM
-
-please use this branch https://github.com/jikunshang/vllm/tree/xpu_kernel to install & test vllm. which already replace rms norm kernel from ipex to vllm-xpu-kernels.
+Please refer to temporary branch https://github.com/jikunshang/vllm/tree/xpu_kernel to install & test vllm which replaces `rms_norm` kernel from IPEX to vllm-xpu-kernels.