diff --git a/.github/workflows/test_api_cuda.yaml b/.github/workflows/test_api_cuda.yaml
index 8b7d18e7..eae4bb8f 100644
--- a/.github/workflows/test_api_cuda.yaml
+++ b/.github/workflows/test_api_cuda.yaml
@@ -24,8 +24,6 @@ jobs:
 
       - name: Run tests
         uses: addnab/docker-run-action@v3
-        env:
-          WORKSPACE: ${{ github.workspace }}
         with:
           image: pytorch/pytorch:2.2.2-cuda12.1-cudnn8-runtime
           options: |
@@ -34,7 +32,7 @@ jobs:
             --shm-size 64G 
             --env USE_CUDA="1" 
             --env PROCESS_SPECIFIC_VRAM="0" 
-            --volume ${{ env.WORKSPACE }}:/workspace
+            --volume ${{ github.workspace }}:/workspace
             --workdir /workspace
           run: |
             pip install -e .[testing,timm,diffusers]
diff --git a/.github/workflows/test_api_rocm.yaml b/.github/workflows/test_api_rocm.yaml
index 5da75559..d472bf13 100644
--- a/.github/workflows/test_api_rocm.yaml
+++ b/.github/workflows/test_api_rocm.yaml
@@ -25,6 +25,11 @@ jobs:
       - name: Checkout code
         uses: actions/checkout@v3
 
+      - name: Target devices
+        run: |
+          echo "DEVICE: $DEVICE"
+          echo "DEVICE=$DEVICE" >> $GITHUB_ENV
+
       - name: Build image
         run: docker build
           --build-arg USER_ID=$(id -u)
@@ -39,7 +44,6 @@ jobs:
         uses: addnab/docker-run-action@v3
         env:
           DEVICE: ${{ env.DEVICE }}
-          WORKSPACE: ${{ github.workspace }}
         with:
           image: opt-bench-rocm:${{ matrix.image.rocm_version }}
           options: |
@@ -49,7 +53,7 @@ jobs:
             --env USE_ROCM="1"
             --device /dev/kfd
             --device /dev/dri/${{ env.DEVICE }}
-            --volume ${{ env.WORKSPACE }}:/workspace
+            --volume ${{ github.workspace }}:/workspace
             --workdir /workspace
           run: |
             pip install -e .[testing,timm,diffusers]
diff --git a/.github/workflows/test_cli_cuda_onnxruntime.yaml b/.github/workflows/test_cli_cuda_onnxruntime.yaml
index bee54f80..c9362244 100644
--- a/.github/workflows/test_cli_cuda_onnxruntime.yaml
+++ b/.github/workflows/test_cli_cuda_onnxruntime.yaml
@@ -21,8 +21,6 @@ jobs:
 
       - name: Run tests
         uses: addnab/docker-run-action@v3
-        env:
-          WORKSPACE: ${{ github.workspace }}
         with:
           image: pytorch/pytorch:2.2.2-cuda11.8-cudnn8-runtime
           options: |
@@ -31,7 +29,7 @@ jobs:
             --shm-size 64G 
             --env USE_CUDA="1" 
             --env PROCESS_SPECIFIC_VRAM="0" 
-            --volume $WORKSPACE:/workspace
+            --volume ${{ github.workspace }}:/workspace
             --workdir /workspace
           run: |
             pip install -e .[testing,onnxruntime-gpu,diffusers,timm] &&
diff --git a/.github/workflows/test_cli_cuda_pytorch_multi_gpu.yaml b/.github/workflows/test_cli_cuda_pytorch_multi_gpu.yaml
index a141f60d..84676ca2 100644
--- a/.github/workflows/test_cli_cuda_pytorch_multi_gpu.yaml
+++ b/.github/workflows/test_cli_cuda_pytorch_multi_gpu.yaml
@@ -24,8 +24,6 @@ jobs:
 
       - name: Run tests
         uses: addnab/docker-run-action@v3
-        env:
-          WORKSPACE: ${{ github.workspace }}
         with:
           image: pytorch/pytorch:2.2.2-cuda12.1-cudnn8-devel
           options: |
@@ -34,7 +32,7 @@ jobs:
             --shm-size 64G
             --env USE_CUDA="1"
             --env PROCESS_SPECIFIC_VRAM="0"
-            --volume $WORKSPACE:/workspace
+            --volume ${{ github.workspace }}:/workspace
             --workdir /workspace
           run: |
             pip install -e .[testing,diffusers,timm,peft,deepspeed]
diff --git a/.github/workflows/test_cli_cuda_pytorch_single_gpu.yaml b/.github/workflows/test_cli_cuda_pytorch_single_gpu.yaml
index d918db0a..8430b45c 100644
--- a/.github/workflows/test_cli_cuda_pytorch_single_gpu.yaml
+++ b/.github/workflows/test_cli_cuda_pytorch_single_gpu.yaml
@@ -24,8 +24,6 @@ jobs:
 
       - name: Run tests
         uses: addnab/docker-run-action@v3
-        env:
-          WORKSPACE: ${{ github.workspace }}
         with:
           image: pytorch/pytorch:2.2.2-cuda12.1-cudnn8-runtime
           options: |
@@ -34,7 +32,7 @@ jobs:
             --shm-size 64G
             --env USE_CUDA="1"
             --env PROCESS_SPECIFIC_VRAM="0"
-            --volume ${{ env.WORKSPACE }}:/workspace
+            --volume ${{ github.workspace }}:/workspace
             --workdir /workspace
           run: |
             pip install -e .[testing,diffusers,timm,peft,bitsandbytes,autoawq,auto-gptq]
diff --git a/.github/workflows/test_cli_cuda_tensorrt_llm.yaml b/.github/workflows/test_cli_cuda_tensorrt_llm.yaml
index 22f4952b..5eef9081 100644
--- a/.github/workflows/test_cli_cuda_tensorrt_llm.yaml
+++ b/.github/workflows/test_cli_cuda_tensorrt_llm.yaml
@@ -24,8 +24,6 @@ jobs:
 
       - name: Run tests
         uses: addnab/docker-run-action@v3
-        env:
-          WORKSPACE: ${{ github.workspace }}
         with:
           image: huggingface/optimum-nvidia:latest
           options: |
@@ -34,7 +32,7 @@ jobs:
             --shm-size 64G 
             --env USE_CUDA="1" 
             --env PROCESS_SPECIFIC_VRAM="0" 
-            --volume ${{ env.WORKSPACE }}:/workspace
+            --volume ${{ github.workspace }}:/workspace
             --workdir /workspace
           run: |
             pip install -e .[testing] &&
diff --git a/.github/workflows/test_cli_cuda_torch_ort_multi_gpu.yaml b/.github/workflows/test_cli_cuda_torch_ort_multi_gpu.yaml
index 1e953002..bef93787 100644
--- a/.github/workflows/test_cli_cuda_torch_ort_multi_gpu.yaml
+++ b/.github/workflows/test_cli_cuda_torch_ort_multi_gpu.yaml
@@ -35,8 +35,6 @@ jobs:
 
       - name: Run tests
         uses: addnab/docker-run-action@v3
-        env:
-          WORKSPACE: ${{ github.workspace }}
         with:
           image: opt-bench-cuda:${{ matrix.image.cuda_version }}
           options: |
@@ -45,7 +43,7 @@ jobs:
             --shm-size 64G
             --env USE_CUDA="1"
             --env PROCESS_SPECIFIC_VRAM="0"
-            --volume ${{ env.WORKSPACE }}:/workspace
+            --volume ${{ github.workspace }}:/workspace
             --workdir /workspace
           run: |
             pip install -e .[testing,torch-ort,deepspeed,peft] &&
diff --git a/.github/workflows/test_cli_cuda_torch_ort_single_gpu.yaml b/.github/workflows/test_cli_cuda_torch_ort_single_gpu.yaml
index fed60cb6..7b4000f6 100644
--- a/.github/workflows/test_cli_cuda_torch_ort_single_gpu.yaml
+++ b/.github/workflows/test_cli_cuda_torch_ort_single_gpu.yaml
@@ -35,8 +35,6 @@ jobs:
 
       - name: Run tests
         uses: addnab/docker-run-action@v3
-        env:
-          WORKSPACE: ${{ github.workspace }}
         with:
           image: opt-bench-cuda:${{ matrix.image.cuda_version }}
           options: |
@@ -45,7 +43,7 @@ jobs:
             --shm-size 64G
             --env USE_CUDA="1"
             --env PROCESS_SPECIFIC_VRAM="0"
-            --volume ${{ env.WORKSPACE }}:/workspace
+            --volume ${{ github.workspace }}:/workspace
             --workdir /workspace
           run: |
             pip install -e .[testing,torch-ort,peft] &&
diff --git a/.github/workflows/test_cli_rocm_pytorch_multi_gpu.yaml b/.github/workflows/test_cli_rocm_pytorch_multi_gpu.yaml
index abe069b1..547fd679 100644
--- a/.github/workflows/test_cli_rocm_pytorch_multi_gpu.yaml
+++ b/.github/workflows/test_cli_rocm_pytorch_multi_gpu.yaml
@@ -25,6 +25,13 @@ jobs:
       - name: Checkout code
         uses: actions/checkout@v3
 
+      - name: Target devices
+        run: |
+          echo "DEVICE0: $DEVICE0"
+          echo "DEVICE1: $DEVICE1"
+          echo "DEVICE0=$DEVICE0" >> $GITHUB_ENV
+          echo "DEVICE1=$DEVICE1" >> $GITHUB_ENV
+
       - name: Build image
         run: docker build
           --build-arg USER_ID=$(id -u)
@@ -40,7 +47,6 @@ jobs:
         env:
           DEVICE0: ${{ env.DEVICE0 }}
           DEVICE1: ${{ env.DEVICE1 }}
-          WORKSPACE: ${{ github.workspace }}
         with:
           image: opt-bench-rocm:${{ matrix.image.rocm_version }}
           options: |
@@ -51,8 +57,8 @@ jobs:
             --device /dev/kfd
             --device /dev/dri/${{ env.DEVICE0 }}
             --device /dev/dri/${{ env.DEVICE1 }}
-            --volume ${{ env.WORKSPACE }}:/workspace
+            --volume ${{ github.workspace }}:/workspace
             --workdir /workspace
           run: |
             pip install -e .[testing,diffusers,timm,deepspeed,peft,autoawq-rocm,auto-gptq-rocm]
-            pytest -x -s -k 'cli and rocm and pytorch and (dp or ddp or device_map or deepspeed)'
+            pytest -x -s -k 'cli and cuda and pytorch and (dp or ddp or device_map or deepspeed)'
diff --git a/.github/workflows/test_cli_rocm_pytorch_single_gpu.yaml b/.github/workflows/test_cli_rocm_pytorch_single_gpu.yaml
index ce00c49f..8f6bb40e 100644
--- a/.github/workflows/test_cli_rocm_pytorch_single_gpu.yaml
+++ b/.github/workflows/test_cli_rocm_pytorch_single_gpu.yaml
@@ -25,6 +25,13 @@ jobs:
       - name: Checkout code
         uses: actions/checkout@v3
 
+      - name: Target devices
+        run: |
+          echo "DEVICE0: $DEVICE0"
+          echo "DEVICE1: $DEVICE1"
+          echo "DEVICE0=$DEVICE0" >> $GITHUB_ENV
+          echo "DEVICE1=$DEVICE1" >> $GITHUB_ENV
+
       - name: Build image
         run: docker build
           --build-arg USER_ID=$(id -u)
@@ -39,7 +46,6 @@ jobs:
         uses: addnab/docker-run-action@v3
         env:
           DEVICE: ${{ env.DEVICE }}
-          WORKSPACE: ${{ github.workspace }}
         with:
           image: opt-bench-rocm:${{ matrix.image.rocm_version }}
           options: |
@@ -49,8 +55,8 @@ jobs:
             --env USE_ROCM="1"
             --device /dev/kfd
             --device /dev/dri/${{ env.DEVICE }}
-            --volume ${{ env.WORKSPACE }}:/workspace
+            --volume ${{ github.workspace }}:/workspace
             --workdir /workspace
           run: |
             pip install -e .[testing,diffusers,timm,peft,autoawq-rocm,auto-gptq-rocm]
-            pytest -x -s -k "cli and rocm and pytorch and not (dp or ddp or device_map or deepspeed) and not (bnb or awq or torch_compile)"
+            pytest -x -s -k "cli and cuda and pytorch and not (dp or ddp or device_map or deepspeed) and not (bnb or awq or torch_compile)"