diff --git a/.github/workflows/self-push.yml b/.github/workflows/self-push.yml
index 210076f1416..ed43d5c02cc 100644
--- a/.github/workflows/self-push.yml
+++ b/.github/workflows/self-push.yml
@@ -5,6 +5,7 @@ on:
     branches:
       - master
       - ci_*
+      - ci-*
     paths:
       - "src/**"
       - "tests/**"
@@ -186,11 +187,85 @@ jobs:
           name: run_all_tests_tf_multi_gpu_test_reports
           path: reports
 
+  run_tests_torch_cuda_extensions_gpu:
+    runs-on: [self-hosted, docker-gpu, single-gpu]
+    container:
+      image: nvcr.io/nvidia/pytorch:21.03-py3
+      options: --gpus 0 --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+    steps:
+      - name: Launcher docker
+        uses: actions/checkout@v2
+
+      - name: NVIDIA-SMI
+        run: |
+          nvidia-smi
+
+      - name: Install dependencies
+        run: |
+          pip install --upgrade pip
+          pip install .[testing,deepspeed]
+
+      - name: Are GPUs recognized by our DL frameworks
+        run: |
+          python -c "import torch; print('Cuda available:', torch.cuda.is_available())"
+          python -c "import torch; print('Cuda version:', torch.version.cuda)"
+          python -c "import torch; print('CuDNN version:', torch.backends.cudnn.version())"
+          python -c "import torch; print('Number of GPUs available:', torch.cuda.device_count())"
+
+      - name: Run all tests on GPU
+        run: |
+          python -m pytest -n 1 --dist=loadfile --make-reports=tests_torch_cuda_extensions_gpu tests/deepspeed tests/extended
+
+      - name: Failure short reports
+        if: ${{ always() }}
+        run: cat reports/tests_torch_cuda_extensions_gpu_failures_short.txt
+
+  run_tests_torch_cuda_extensions_multi_gpu:
+    runs-on: [self-hosted, docker-gpu, multi-gpu]
+    container:
+      image: nvcr.io/nvidia/pytorch:21.03-py3
+      options: --gpus 0 --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+    steps:
+      - name: Launcher docker
+        uses: actions/checkout@v2
+
+      - name: NVIDIA-SMI
+        run: |
+          nvidia-smi
+
+      - name: Install dependencies
+        run: |
+          pip install --upgrade pip
+          pip install .[testing,deepspeed,fairscale]
+
+      - name: Are GPUs recognized by our DL frameworks
+        run: |
+          python -c "import torch; print('Cuda available:', torch.cuda.is_available())"
+          python -c "import torch; print('Cuda version:', torch.version.cuda)"
+          python -c "import torch; print('CuDNN version:', torch.backends.cudnn.version())"
+          python -c "import torch; print('Number of GPUs available:', torch.cuda.device_count())"
+
+      - name: Run all tests on GPU
+        run: |
+          python -m pytest -n 1 --dist=loadfile --make-reports=tests_torch_cuda_extensions_multi_gpu tests/deepspeed tests/extended
+
+      - name: Failure short reports
+        if: ${{ always() }}
+        run: cat reports/tests_torch_cuda_extensions_multi_gpu_failures_short.txt
+
+
   send_results:
     name: Send results to webhook
     runs-on: ubuntu-latest
     if: always()
-    needs: [run_tests_torch_gpu, run_tests_tf_gpu, run_tests_torch_multi_gpu, run_tests_tf_multi_gpu]
+    needs: [
+        run_tests_torch_gpu,
+        run_tests_tf_gpu,
+        run_tests_torch_multi_gpu,
+        run_tests_tf_multi_gpu,
+        run_tests_torch_cuda_extensions_gpu,
+        run_tests_torch_cuda_extensions_multi_gpu
+    ]
     steps:
       - uses: actions/checkout@v2
 
diff --git a/.github/workflows/self-scheduled.yml b/.github/workflows/self-scheduled.yml
index 978d9e02a69..df9148c38e0 100644
--- a/.github/workflows/self-scheduled.yml
+++ b/.github/workflows/self-scheduled.yml
@@ -246,11 +246,84 @@ jobs:
           name: run_all_tests_tf_multi_gpu_test_reports
           path: reports
 
+  run_all_tests_torch_cuda_extensions_gpu:
+    runs-on: [self-hosted, docker-gpu, single-gpu]
+    container:
+      image: nvcr.io/nvidia/pytorch:21.03-py3
+      options: --gpus 0 --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+    steps:
+      - name: Launcher docker
+        uses: actions/checkout@v2
+
+      - name: NVIDIA-SMI
+        run: |
+          nvidia-smi
+
+      - name: Install dependencies
+        run: |
+          pip install --upgrade pip
+          pip install .[testing,deepspeed]
+
+      - name: Are GPUs recognized by our DL frameworks
+        run: |
+          python -c "import torch; print('Cuda available:', torch.cuda.is_available())"
+          python -c "import torch; print('Cuda version:', torch.version.cuda)"
+          python -c "import torch; print('CuDNN version:', torch.backends.cudnn.version())"
+          python -c "import torch; print('Number of GPUs available:', torch.cuda.device_count())"
+
+      - name: Run all tests on GPU
+        run: |
+          python -m pytest -n 1 --dist=loadfile --make-reports=tests_torch_cuda_extensions_gpu tests/deepspeed tests/extended
+
+      - name: Failure short reports
+        if: ${{ always() }}
+        run: cat reports/tests_torch_cuda_extensions_gpu_failures_short.txt
+
+  run_all_tests_torch_cuda_extensions_multi_gpu:
+    runs-on: [self-hosted, docker-gpu, multi-gpu]
+    container:
+      image: nvcr.io/nvidia/pytorch:21.03-py3
+      options: --gpus 0 --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+    steps:
+      - name: Launcher docker
+        uses: actions/checkout@v2
+
+      - name: NVIDIA-SMI
+        run: |
+          nvidia-smi
+
+      - name: Install dependencies
+        run: |
+          pip install --upgrade pip
+          pip install .[testing,deepspeed,fairscale]
+
+      - name: Are GPUs recognized by our DL frameworks
+        run: |
+          python -c "import torch; print('Cuda available:', torch.cuda.is_available())"
+          python -c "import torch; print('Cuda version:', torch.version.cuda)"
+          python -c "import torch; print('CuDNN version:', torch.backends.cudnn.version())"
+          python -c "import torch; print('Number of GPUs available:', torch.cuda.device_count())"
+
+      - name: Run all tests on GPU
+        run: |
+          python -m pytest -n 1 --dist=loadfile --make-reports=tests_torch_cuda_extensions_multi_gpu tests/deepspeed tests/extended
+
+      - name: Failure short reports
+        if: ${{ always() }}
+        run: cat reports/tests_torch_cuda_extensions_multi_gpu_failures_short.txt
+
   send_results:
     name: Send results to webhook
     runs-on: ubuntu-latest
     if: always()
-    needs: [run_all_tests_torch_gpu, run_all_tests_tf_gpu, run_all_tests_torch_multi_gpu, run_all_tests_tf_multi_gpu]
+    needs: [
+        run_all_tests_torch_gpu,
+        run_all_tests_tf_gpu,
+        run_all_tests_torch_multi_gpu,
+        run_all_tests_tf_multi_gpu,
+        run_all_tests_torch_cuda_extensions_gpu,
+        run_all_tests_torch_cuda_extensions_multi_gpu
+    ]
     steps:
       - uses: actions/checkout@v2