diff --git a/release/long_running_distributed_tests/autoscaler-cluster.yaml b/release/long_running_distributed_tests/autoscaler-cluster.yaml
index 7bb1f05c1..9373786bd 100644
--- a/release/long_running_distributed_tests/autoscaler-cluster.yaml
+++ b/release/long_running_distributed_tests/autoscaler-cluster.yaml
@@ -5,11 +5,6 @@ max_workers: 3
 
 idle_timeout_minutes: 15
 
-docker:
-    image: rayproject/ray-ml:1.2.0-gpu
-    container_name: ray_container
-    pull_before_run: True
-
 provider:
     type: aws
     region: us-west-2
@@ -21,17 +16,16 @@ auth:
 
 head_node:
     InstanceType: g3.8xlarge
-    ImageId: ami-0828a1066dc750737
+    ImageId: ami-0b1a80ce62c464a55
 
 worker_nodes:
   InstanceType: g3.8xlarge
-  ImageId: ami-0828a1066dc750737
-  InstanceMarketOptions:
-    MarketType: spot
+  ImageId: ami-0b1a80ce62c464a55
 
 setup_commands:
-  - sudo apt-get install -y libglib2.0-0 libcudnn7=7.6.5.32-1+cuda10.1
+#  - sudo apt-get install -y libglib2.0-0 libcudnn7=7.6.5.32-1+cuda10.1
   - pip install -U https://ray-wheels.s3-us-west-2.amazonaws.com/releases/1.2.0/4c71f76b25bde02f208bfb21451347834994a72a/ray-1.2.0-cp37-cp37m-manylinux2014_x86_64.whl
+  - pip install torch==1.4.0 torchvision==0.5.0 ray[all]
 
 # Command to start ray on the head node. You don't need to change this.
 head_start_ray_commands: