diff --git a/README.md b/README.md
index 3033082..b23c8d0 100644
--- a/README.md
+++ b/README.md
@@ -45,7 +45,7 @@ For model training the data provided by [UniProt](https://www.uniprot.org/) is u
 You can install the requirements with the following
 
 ```bash
-$ python setup.py install --user
+$ pip install .
 ```
 
 Then, you must install Microsoft's sparse attention CUDA kernel with the following two steps.
diff --git a/install_deepspeed.sh b/install_deepspeed.sh
index 79add81..928d21e 100644
--- a/install_deepspeed.sh
+++ b/install_deepspeed.sh
@@ -1,3 +1,3 @@
 sudo apt-get -y install llvm-9-dev cmake
 git clone https://github.com/microsoft/DeepSpeed.git /tmp/Deepspeed
-cd /tmp/Deepspeed && DS_BUILD_SPARSE_ATTN=1 ./install.sh -s
+cd /tmp/Deepspeed && DS_BUILD_SPARSE_ATTN=1 ./install.sh