Add MII tests (#2533)

Adding MII tests to ensure changes to DS-Inference do not break MII

.github/workflows/nv-mii.yml

+name: nv-mii
+
+on:
+  push:
+    branches:
+      - 'master'
+      - 'staging**'
+    paths-ignore:
+      - 'docs/**'
+  pull_request:
+    paths-ignore:
+      - 'docs/**'
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  unit-tests:
+    runs-on: [self-hosted, nvidia, cu116, v100]
+
+    steps:
+      - uses: actions/checkout@v2
+
+      - name: environment
+        run: |
+          echo "JobID: $AISC_NODE_INSTANCE_ID"
+          nvidia-smi
+          which python
+          python --version
+          which nvcc
+          nvcc --version
+          pip install --upgrade pip
+          pip uninstall --yes torch torchvision triton
+          pip install torch torchvision --extra-index-url https://download.pytorch.org/whl/cu116
+          python -c "import torch; print('torch:', torch.__version__, torch)"
+          python -c "import torch; print('CUDA available:', torch.cuda.is_available())"
+
+      - name: Install MII
+        run: |
+          pip uninstall --yes deepspeed deepspeed-mii transformers
+          pip install .[dev]
+          pip install git+https://github.com/huggingface/transformers.git
+
+      - name: Python environment
+        run: |
+          pip list
+
+      - name: Unit tests
+        run: |
+          git clone https://github.com/microsoft/DeepSpeed-MII.git
+          cd DeepSpeed-MII
+          pip install .[dev]
+          unset TORCH_CUDA_ARCH_LIST # only jit compile for current arch
+          if [[ -d ./torch-extensions ]]; then rm -rf ./torch-extensions; fi
+          cd tests
+          TRANSFORMERS_CACHE=/blob/transformers_cache/ TORCH_EXTENSIONS_DIR=./torch-extensions pytest --color=yes --durations=0 --forked --verbose -m "CPU or local" ./

deepspeed/inference/config.py

@@ -233,7 +233,9 @@ class DeepSpeedInferenceConfig(DeepSpeedConfigModel):
     injection_policy_tuple: tuple = None
     """ TODO: Add docs """
 
-    config: Dict = None  # todo: really no need for this field if we can refactor
+    config: Dict = Field(
+        None,
+        alias="args")  # todo: really no need for this field if we can refactor
 
     max_out_tokens: int = Field(1024, alias="max_tokens")
     """