diff --git a/.build/Dockerfile b/.build/Dockerfile
index 0328c86..c007c06 100644
--- a/.build/Dockerfile
+++ b/.build/Dockerfile
@@ -1,6 +1,6 @@
 # Use NVIDIA CUDA as base image and run the same installation as in the other packages.
 # The version of cudatoolkit must match those of the base image, see Dockerfile.pytorch
-FROM nvidia/cuda:10.2-base-ubuntu18.04
+FROM nvidia/cuda:10.1-base-ubuntu18.04
 LABEL maintainer="Jupyter Project <jupyter@googlegroups.com>"
 
 ############################################################################
@@ -380,7 +380,7 @@ RUN conda install --quiet --yes \
 
 # Install PyTorch, version of cudatoolkit must match those of the base image
 RUN conda install -y -c pytorch \
-    cudatoolkit=10.2 \
+    cudatoolkit=10.1 \
     'pytorch=1.3.1' \
     torchvision && \
     conda clean --all -f -y && \
diff --git a/README.md b/README.md
index 0c01c24..aec1668 100644
--- a/README.md
+++ b/README.md
@@ -75,7 +75,7 @@ In order to stop the local deployment, run:
  
 A Jupyter instance often requires data from other services. 
 If that data-source is containerized in Docker and sharing a port for communication shouldn't be allowed, e.g., for security reasons,
-then connecting the data-source with *GPU-Jupyter* within a Docker Swarm is a great option! \
+then connecting the data-source with *GPU-Jupyter* within a Docker Swarm is a great option! 
 
 ### Set up Docker Swarm and Registry
 
diff --git a/docker-compose.yml b/docker-compose.yml
index 66970f7..e8f8351 100755
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -1,7 +1,6 @@
 version: "3.4"
 services:
   gpu-jupyter:
-    image: 127.0.0.1:5001/gpu-jupyter
     build: .build
     ports:
       - ${JUPYTER_PORT}:8888