[fbsync] [Nova] Add GHA Linux CPU Unittests for Torchvision (#6759)

Summary:
* [Nova][WIP] Add Linux CPU Unittests for Torchvision

* use conda-builder image since conda installation is needed

* install torch dep with conda instead

* use circleCI command to run tests

* larger instance to avoid OOM issues

* proper syntax for self-hosted runners

* 4xlarge instance

* 8xlarge

* 12xlarge

* use setup-miniconda job

* add back PATH change to help setup py detect conda

* run conda shell script

* install other deps up front

* git config and undo path change

* revert to local conda install

* conda-builder image

* support for whole python version matrix

* clean up the conda env once we are done with the job

Reviewed By: YosuaMichael

Differential Revision: D40588169

fbshipit-source-id: 515b12daa84d1707f6b700782fade13f8532ff05

Author: Joao Gomes

.github/workflows/test-linux-cpu.yml

@@ -0,0 +1,69 @@
+name: Unit-tests on Linux CPU
+
+on:
+  pull_request:
+  push:
+    branches:
+      - nightly
+      - main
+      - release/*
+  workflow_dispatch:
+
+env:
+  CHANNEL: "nightly"
+
+jobs:
+  tests:
+    name: "Unit-tests on Linux CPU"
+    runs-on: [self-hosted, linux.12xlarge]
+    container:
+      image: pytorch/conda-builder:cpu
+    strategy:
+      matrix:
+        py_vers: ["3.7", "3.8", "3.9", "3.10"]
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v2
+      - name: Set Release CHANNEL (for release)
+        if: ${{ (github.event_name == 'pull_request' && startsWith(github.base_ref, 'release')) || startsWith(github.ref, 'refs/heads/release') }}
+        run: |
+          echo "CHANNEL=test" >> "$GITHUB_ENV"
+      - name: Setup Conda
+        shell: bash -l {0}
+        env:
+          ENV_NAME: conda-env-${{ github.run_id }}
+          PY_VERS: ${{ matrix.py_vers }}
+        run: |
+          git config --global --add safe.directory /__w/vision/vision
+          . ~/miniconda3/etc/profile.d/conda.sh
+          conda create -yp ${ENV_NAME} python=${PY_VERS} numpy libpng jpeg scipy
+          echo "CONDA_RUN=conda run -p ${ENV_NAME}" >> "$GITHUB_ENV"
+      - name: Install TorchVision
+        shell: bash -l {0}
+        env:
+          VERSION: cpu
+          CUDATOOLKIT: cpuonly
+        run: |
+          # Needed for JPEG library detection as setup.py detects conda presence
+          # by running `shutil.which('conda')`
+          export PATH=~/miniconda3/bin:$PATH
+          set -ex
+          ${CONDA_RUN} conda install \
+            --yes \
+            -c "pytorch-${CHANNEL}" \
+            -c nvidia "pytorch-${CHANNEL}"::pytorch[build="*${VERSION}*"] \
+            "${CUDATOOLKIT}"
+          ${CONDA_RUN} python3 setup.py develop
+          ${CONDA_RUN} python3 -m pip install pytest pytest-mock av
+      - name: Run tests
+        shell: bash -l {0}
+        env:
+          ENV_NAME: conda-env-${{ github.run_id }}
+          PY_VERS: ${{ matrix.py_vers }}
+        run: |
+          . ~/miniconda3/etc/profile.d/conda.sh
+          set -ex
+          ${CONDA_RUN} python3 -m torch.utils.collect_env
+          ${CONDA_RUN} python3 -m pytest --junitxml=test-results/junit.xml -v --durations 20
+          conda env remove -p ${ENV_NAME}