diff --git a/.github/workflows/docker.yml b/.github/workflows/docker.yml
index 71a4e59..e64b3ba 100644
--- a/.github/workflows/docker.yml
+++ b/.github/workflows/docker.yml
@@ -257,15 +257,15 @@ jobs:
           alias_image torch-cuda116 latest torch latest
           push_image
 
-  qpod_torch_cuda102:
-    name: qpod/torch-cuda102
-    needs: qpod_cuda_10_2
+  qpod_torch_cuda112:
+    name: qpod/torch-cuda112
+    needs: qpod_cuda_11_2
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3
       - run: |
           source ./tool.sh
-          build_image torch-cuda102 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_10.2" --build-arg "ARG_PROFILE_PYTHON=torch"
+          build_image torch-cuda112 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_11.2" --build-arg "ARG_PROFILE_PYTHON=torch"
           push_image
 
   qpod_paddle_cuda116:
@@ -280,21 +280,21 @@ jobs:
           alias_image paddle-cuda116 latest paddle latest
           push_image
 
-  qpod_paddle_cuda102:
-    name: qpod/paddle-cuda102
-    needs: qpod_cuda_10_2
+  qpod_paddle_cuda112:
+    name: qpod/paddle-cuda112
+    needs: qpod_cuda_11_2
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3
       - run: |
           source ./tool.sh
-          build_image paddle-cuda102 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_10.2" --build-arg "ARG_PROFILE_PYTHON=paddle,mkl"
+          build_image paddle-cuda112 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_11.2" --build-arg "ARG_PROFILE_PYTHON=paddle,mkl"
           push_image
 
 
   qpod_py-nlp-cuda116:
     name: qpod/py-nlp-cuda116
-    needs: qpod_base
+    needs: qpod_cuda_11_6
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3
@@ -306,7 +306,7 @@ jobs:
 
   qpod_py-nlp-cuda112:
     name: qpod/py-nlp-cuda112
-    needs: qpod_base
+    needs: qpod_cuda_11_2
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3
@@ -315,21 +315,10 @@ jobs:
           build_image py-nlp-cuda112 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_11.2" --build-arg "ARG_PROFILE_PYTHON=datascience,mkl,torch,nlp"
           push_image
 
-  qpod_py-nlp-cuda102:
-    name: qpod/py-nlp-cuda102
-    needs: qpod_base
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v3
-      - run: |
-          source ./tool.sh
-          build_image py-nlp-cuda102 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_10.2" --build-arg "ARG_PROFILE_PYTHON=datascience,mkl,torch,nlp"
-          push_image
-
 
   qpod_py-cv:
     name: qpod/py-cv
-    needs: qpod_base
+    needs: qpod_cuda_11_6
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3
diff --git a/docker_cuda/Dockerfile b/docker_cuda/Dockerfile
index 279ae2e..6001520 100644
--- a/docker_cuda/Dockerfile
+++ b/docker_cuda/Dockerfile
@@ -7,7 +7,7 @@ FROM ${BASE_NAMESPACE:+$BASE_NAMESPACE/}${BASE_IMG}
 LABEL maintainer="haobibo@gmail.com"
 
 # Let NVIDIA docker ignore cuda requirement check
-ENV NVIDIA_DISABLE_REUIRE=1
+ENV NVIDIA_DISABLE_REQUIRE=1
 
 # For cuda version 10.0, the image is solely serverd for legacy tensorflow 1.15, which requires python 3.7
 # For tensorflow 2.x or torch, python>=3.9 is supported.