diff --git a/.ci/docker/common/install_pip_requirements.sh b/.ci/docker/common/install_pip_requirements.sh
index a548d200462..acd02da24ab 100644
--- a/.ci/docker/common/install_pip_requirements.sh
+++ b/.ci/docker/common/install_pip_requirements.sh
@@ -4,4 +4,4 @@ set -ex
 
 # Install pip packages
 pip install --upgrade pip
-pip install -r ./requirements.txt
+pip install --extra-index-url https://pypi.nvidia.com -r ./requirements.txt
diff --git a/beginner_source/basics/tensorqs_tutorial.py b/beginner_source/basics/tensorqs_tutorial.py
index 30e05cb10d0..55d8b62f084 100644
--- a/beginner_source/basics/tensorqs_tutorial.py
+++ b/beginner_source/basics/tensorqs_tutorial.py
@@ -44,6 +44,8 @@
 # Tensors can be created from NumPy arrays (and vice versa - see :ref:`bridge-to-np-label`).
 np_array = np.array(data)
 x_np = torch.from_numpy(np_array)
+# Note: torch.from_numpy shares memory with the NumPy array when possible. If you want a copy, use torch.tensor(np_array).
+# torch.as_tensor(np_array) also avoids a copy when possible.
 
 
 ###############################################################