diff --git a/docs/tutorials/deploy/deploy_server/deploy_cpp/deploy_cpp_win_vs2019.md b/docs/tutorials/deploy/deploy_server/deploy_cpp/deploy_cpp_win_vs2019.md
index a1b659cb65db1d6774e1797732054dceef590711..4cf0a7f55b43e4cda7a583c790e86e2935092325 100755
--- a/docs/tutorials/deploy/deploy_server/deploy_cpp/deploy_cpp_win_vs2019.md
+++ b/docs/tutorials/deploy/deploy_server/deploy_cpp/deploy_cpp_win_vs2019.md
@@ -29,7 +29,7 @@ git clone https://github.com/PaddlePaddle/PaddleX.git
 
 PaddlePaddle C++ 预测库针对不同的`CPU`，`CUDA`，以及是否支持TensorRT，提供了不同的预编译版本，目前PaddleX依赖于Paddle1.8版本，以下提供了多个不同版本的Paddle预测库:
 
-|  版本说明   | 预测库(1.8.2版本)  | 编译器 | 构建工具| cuDNN | CUDA
+|  版本说明   | 预测库(1.8.2版本)  | 编译器 | 构建工具| cuDNN | CUDA |
 |  ----  |  ----  |  ----  |  ----  | ---- | ---- |
 | cpu_avx_mkl  | [fluid_inference.zip](https://paddle-wheel.bj.bcebos.com/1.8.2/win-infer/mkl/cpu/fluid_inference_install_dir.zip) | MSVC 2015 update 3 | CMake v3.16.0 |
 | cpu_avx_openblas  | [fluid_inference.zip](https://paddle-wheel.bj.bcebos.com/1.8.2/win-infer/open/cpu/fluid_inference_install_dir.zip) | MSVC 2015 update 3 | CMake v3.16.0 |