super-gradients/.circleci/config.yml

version: 2.1

parameters:
  ad_hoc_container_build:
    type: boolean
    default: false
  ad_hoc_container_build_code_only:
    type: boolean
    default: false
  remote_docker_version:
    type: string
    description: remote docker version
    default: "20.10.18"
  sg_docker_version:
    type: string
    description: the version of super gradients docker image
    default: "2.0.0"
  build_py_version:
    type: string
    description: Python version
    default: "3.8"
  run_sanity_tests_flow:
    type: boolean
    description: If true will run sanity_test flow
    default: false
  skip_md_files:
    type: boolean
    default: true
  orb_version:
    type: string
    description: Deci ai ORB version https://circleci.com/developer/orbs/orb/deci-ai/circleci-common-orb
    default: "10.5.1"
#    default: "dev:alpha"

orbs:
  deci-common: deci-ai/circleci-common-orb@<< pipeline.parameters.orb_version >>
  aws-ecr: circleci/aws-ecr@8.2.1
  docker: circleci/docker@2.2.0
  slack: circleci/slack@4.12.0 # see https://circleci.com/developer/orbs/orb/circleci/slack for examples and more

# This filter operates on SemVer2 tags only
release_tag_filter: &release_tag_filter
  filters:
    branches:
      ignore: /.*/
    tags:
      only: /^\d+\.\d+\.\d+$/

release_candidate_filter: &release_candidate_filter
  filters:
    branches:
      # Apply to master branch and other branches that start with "master"
      only: /^master.*/

release_candidate_tag_filter: &release_candidate_tag_filter
  filters:
    branches:
      ignore: /.*/
    tags:
      only: /^\d+\.\d+\.\d+rc\d+/

commands:
  build_and_publish_command:
    parameters:
      repo_name:
        type: string
      docker_context:
        type: string
      image_tag:
        type: string
      additional_tags:
        type: string
      build_args:
        type: string
        default: ""
      dockerfile:
        type: string
        default: "Dockerfile"
    steps:
      - checkout
      - attach_workspace:
          at: ~/
      - run:
          name: Put config dir in repo context
          command: |
            if [ -d ~/.config ]; then
              echo "found a .config directory, copying to repo dir"
              cp -r  ~/.config ~/project/<< parameters.docker_context >>
            fi
      - deci-common/ecr_login_dev
      - deci-common/container_image_build:
          context: << parameters.docker_context >>
          working_directory: "."
          repository_name: << parameters.repo_name >>
          image_tag: << parameters.image_tag >>
          dockerfile: << parameters.dockerfile >>
          build_args: << parameters.build_args >>
      #          build_args: "PYTHON_VERSION=3.8 SG_VERSION=3.0.7"
      - deci-common/push_docker_image_aws_dev:
          repository_name: << parameters.repo_name >>
          image_tag: << parameters.image_tag >>
          additional_tags: << parameters.additional_tags >>

  get_beta_and_rc_tags:
    description: "getting beta and rc tag (if exist) according to ouir convention"
    steps:
        - attach_workspace:
           at: ~/
        - run:
           name: push BETA_TAG and RC_TAG variable to BASH_ENV
           command: |
              if [[ -f ~/BETA_TAG ]]; then
                echo 'export BETA_TAG="$(cat ~/BETA_TAG)"' >> "$BASH_ENV"
                source $BASH_ENV
              fi

              if [[ -f ~/RC_TAG ]]; then
                echo 'export RC_TAG="$(cat ~/RC_TAG)"' >> "$BASH_ENV"
                source $BASH_ENV
              fi 
              echo "RC_TAG=$RC_TAG   ||  BETA_TAG=$BETA_TAG"

  adding_tag_to_ecr_container_image:
   description: adding a tag to an existing  container image
   parameters:
    image_repo:
      type: string
    new_image_tag:
      type: string
    source_image_tag:
      type: string
   steps:
     - deci-common/run_on_dev_account:
           command: |
              MANIFEST=$(aws ecr batch-get-image --repository-name << parameters.image_repo >> --image-ids imageTag=<< parameters.source_image_tag >> --query 'images[].imageManifest' --output text)
              aws ecr put-image --repository-name << parameters.image_repo >> --image-tag << parameters.new_image_tag >> --image-manifest "$MANIFEST"
              echo "added tag: << parameters.new_image_tag >>    to image:  << parameters.image_repo >>:<< parameters.new_image_tag >>"

jobs:
  build:
    environment:
      CIRCLE_COMPARE_URL: << pipeline.project.git_url >>/compare/<< pipeline.git.base_revision >>..<<pipeline.git.revision>>
    parameters:
      py_version:
        type: string
        default: latest
      package_name:
        type: string
    docker:
      - image: cimg/python:<< parameters.py_version >>
    resource_class: large
    steps:
      - deci-common/checkout_and_skip_build:
          check_version_file: true
          skip_md_files: << pipeline.parameters.skip_md_files >>
      - deci-common/get_persisted_version_info
      - unless:
          condition:
            equal: [ master, << pipeline.git.branch >> ]
          steps:
            - run:
                name: install Black Flake8 python linter
                command: |
                  pip install --user -r requirements.dev.txt
            - run:
                name: Lint all python files changed since develop branch
                command: |
                  flake8 --statistics --config scripts/flake8-config setup.py $(git diff --diff-filter ACM origin/master --name-only | grep 'py$' | grep -v 'experimental/' | grep -v 'experimental_models/')
            - run:
                name: Run Black on changed files against master branch
                command: |
                  black --check setup.py $(git diff --diff-filter ACM origin/master --name-only | grep 'py$' | grep -v 'experimental/' | grep -v 'experimental_models/')
      - run:
          name: add requirements.txt and requirements.pro.txt to source code
          command: |
            cp requirements.txt src/super_gradients/requirements.txt
            cp requirements.pro.txt src/super_gradients/requirements.pro.txt
      - run:
          name: install python dependencies
          command: |
            python3 -m venv venv
            . venv/bin/activate
            python3 -m pip install pip==23.1.2
            python3 -m pip install -r requirements.txt
      - run:
          name: edit package version
          command: |
            echo "${NEW_VERSION}" > version.txt
            cat version.txt
      - run:
          name: setup custom environment variables
          command: |
            echo 'export UPLOAD_LOGS=FALSE' >> $BASH_ENV
      - run:
          name: install package
          no_output_timeout: 30m
          command: |
            . venv/bin/activate
            python3 -m pip uninstall -y super_gradients
            python3 -m pip install -e .[pro] --extra-index-url https://pypi.ngc.nvidia.com

      - run:
          name: install pytorch quantization package
          no_output_timeout: 30m
          command: |
            . venv/bin/activate
            python3 -m pip install pytorch-quantization==2.1.2 --extra-index-url https://pypi.ngc.nvidia.com
            python3 -m pip install onnx_graphsurgeon==0.3.27   --extra-index-url https://pypi.ngc.nvidia.com
      - run:
          name: run tests with coverage
          no_output_timeout: 30m
          command: |
            . venv/bin/activate
            coverage run --source=super_gradients -m unittest tests/deci_core_unit_test_suite_runner.py
            coverage report
            coverage html  # open htmlcov/index.html in a browser

      - store_artifacts:
          path: htmlcov

      - store_artifacts:
          path: ~/sg_logs

  breaking_change_check:
    parameters:
      py_version:
        type: string
        default: latest
    docker:
      - image: cimg/python:<< parameters.py_version >>
    resource_class: small
    steps:
      - checkout
      - run:
          name: Detect breaking changes
          no_output_timeout: 30m
          command: |
            python3 -m venv venv
            . venv/bin/activate
            python3 -m pip install pip==23.1.2
            python3 -m pip install -e .
            python3 -m pip install gitpython==3.1.0
            coverage run --source=super_gradients -m unittest tests/breaking_change_tests/unit_test.py
            coverage run --source=super_gradients -m unittest tests/breaking_change_tests/test_detect_breaking_change.py
      - run:
          name: Remove new environment when failed
          command: "rm -r venv"
          when: on_fail
      - slack/notify:
          channel: "sg-integration-tests"
          event: fail
          template: basic_fail_1 # see https://github.com/CircleCI-Public/slack-orb/wiki#templates.

  change_rc_to_b:
     description: "change rc in the tag to b"
     docker:
       - image: cimg/base:stable-20.04
     resource_class: small
     steps:
       - run:
          name: change the rc to b
          command:
           |
            BETA_TAG=$(echo ${CIRCLE_TAG} | sed -e  's|rc|b|'); echo ${BETA_TAG} >> ~/BETA_TAG
            cat ~/BETA_TAG
       - persist_to_workspace:
           root: ~/
           paths:
             - "BETA_TAG"




  build_and_publish_sg_container:
    description : "building the temp image and pushing to dev ecr"
    parameters:
       repo_name:
         type: string
         default: 'deci/super-gradients'
       sg_python_version:
         type: string
         default: $CIRCLE_TAG
    docker:
       - image: cimg/base:stable-20.04
    resource_class: small
    steps:
     - checkout
     - get_beta_and_rc_tags
     - run:
         command: echo *$BETA_TAG*
     - setup_remote_docker:
          version: << pipeline.parameters.remote_docker_version >>
          docker_layer_caching: true
     - deci-common/container_image_build:
           dockerfile: scripts/Dockerfile
           repository_name: << parameters.repo_name >>
           image_tag: "$BETA_TAG"
           build_args: 'SG_VERSION=<< parameters.sg_python_version >>'
           extra_flags: ' --compress'
     - deci-common/push_docker_image_aws_dev:
           repository_name: << parameters.repo_name >>
           image_tag: "$BETA_TAG"
           additional_tags: "latest"

  find_rc_tag_per_sha:
   description: this command will take the sha of the last commit and find the rc tag it belongs to as ${RC_TAG} variale
   docker:
      - image: cimg/aws:2022.11.1
   resource_class: small
   steps:
     - checkout
     - run:
          name: get rc tag of the final tag
          command: |
              if [[ $(RC_TAG=$(git tag --contains ${CIRCLE_SHA1} | grep -E '[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+rc[[:digit:]]+' | head -n 1)) ]] ; then
                echo "Found RC version: ${RC_TAG}"
                echo $RC_TAG >> ~/RC_TAG
              else
               echo "No Rc Tag found for commit"
               circleci-agent step halt
              fi

     - persist_to_workspace:
           root: ~/
           paths:
             - "RC_TAG"


  add_rc_tag_to_beta:
   description: in the event of successful test promote beta to rc
   parameters:
     image_repo:
        type: string
        default: "deci/super-gradients"
   docker:
       - image: cimg/base:stable-20.04
   resource_class: small
   steps:
     - get_beta_and_rc_tags
     - run:
         command: |
           echo $BETA_TAG
     - adding_tag_to_ecr_container_image:
        image_repo: << parameters.image_repo >>
        source_image_tag: $BETA_TAG
        new_image_tag: $CIRCLE_TAG

  add_release_tag_to_rc:
    description: in the event of successful test promote rc to release and latest
    parameters:
      image_repo:
        type: string
        default: "deci/super-gradients"
    docker:
      - image: cimg/base:stable-20.04
    resource_class: small
    steps:
      - get_beta_and_rc_tags
      - run:
          command: |
            if [ -z $RC_TAG ] ; then
              circleci-agent step halt
            else
              echo $RC_TAG
            fi
      - adding_tag_to_ecr_container_image:
          image_repo: << parameters.image_repo >>
          source_image_tag: $RC_TAG
          new_image_tag: ${CIRCLE_TAG}

  testing_supergradients_docker_image:
    description: "running integration test on the code"
    parameters:
      image_repo:
        type: string
    #reserved for testing when will be done
    #machine: true
    #resource_class: deci-ai/research-standard
    docker:
      - image: cimg/base:stable-20.04
    steps:
        - deci-common/ecr_login_dev
        - get_beta_and_rc_tags
        # - run:
        #    name: integration test
        #    #command: sudo docker run -it -e ENVIRONMENT_NAME=production -v ${PWD}:/SG -v /data:/data deciai/super-gradients:3.0.0 python3 ./tests/integration_tests/pretrained_models_test.py
        #    command: docker run --rm -it --shm-size=2gb --gpus all -v ${PWD}:/SG -v /data:/data << parameters.image_repo >>:<< parameters.image_tag >> python3 -c 'print("it works!")'
        -  run:
             command: echo Hello world


  release_candidate:
    environment:
      CIRCLE_COMPARE_URL: << pipeline.project.git_url >>/compare/<< pipeline.git.base_revision >>..<<pipeline.git.revision>>
    parameters:
      py_version:
        type: string
    docker:
      - image: cimg/python:<< parameters.py_version >>
    steps:
      - deci-common/checkout_and_skip_build:
          check_version_file: true
          skip_md_files: true

      - deci-common/get_persisted_version_info
      - run:
          name: edit package version
          command: |
            echo $NEW_VERSION > version.txt
      - deci-common/pip_upload_package_codeartifact_dev:
          codeartifact_repository: "deci-packages"
      - deci-common/pip_test_package_installation_codeartifact_dev:
          package_name: "super-gradients"
          version: $NEW_VERSION
      - deci-common/git_config_automation_user
      - run:
          name: "commit version file"
          command: |
            git commit version.txt -m "Deci Services - Changed version to $NEW_VERSION"
      - deci-common/git_commit_and_tag:
          version: $NEW_VERSION

  release_version:
    environment:
      CIRCLE_COMPARE_URL: << pipeline.project.git_url >>/compare/<< pipeline.git.base_revision >>..<<pipeline.git.revision>>
    parameters:
      py_version:
        type: string
      dev_venv_name:
        type: string
        default: "dev-sg-${CIRCLE_BUILD_NUM}"
    docker:
      - image: cimg/python:<< parameters.py_version >>
    steps:
      - deci-common/checkout_and_skip_build:
          check_version_file: true
          skip_md_files: false
      - run:
          name: add requirements.txt and requirements.pro.txt to source code
          command: |
            cp requirements.txt src/super_gradients/requirements.txt
            cp requirements.pro.txt src/super_gradients/requirements.pro.txt
      - run:
          name: edit package version
          command: |
            echo $CIRCLE_TAG > version.txt

      - deci-common/pip_upload_package_codeartifact_all_accounts:
          codeartifact_repository: "deci-packages"

      - deci-common/pip_test_package_installation_codeartifact_dev:
          package_name: "super-gradients"
          version: $CIRCLE_TAG
          venv_name: << parameters.dev_venv_name >>
      - run:
          name: verify that the output of __version__ is what we expect
          command: |
            . << parameters.dev_venv_name >>-super-gradients-$CIRCLE_TAG/bin/activate
            python3 tests/verify_version.py $CIRCLE_TAG

      - deci-common/pip_test_package_installation_codeartifact_prod:
          package_name: "super-gradients"
          version: $CIRCLE_TAG

      - deci-common/pip_upload_package_codeartifact_prod:
          codeartifact_repository: "deci-toolkit"

      - deci-common/git_commit_and_tag:
          version: $CIRCLE_TAG
          delete_remote_tag_before_tagging: true

      - deci-common/tag_as:
          tag_name: "stable"
          delete_remote: true

      - deci-common/github_create_release:
          github_cli_token: $GITHUB_CLI_TOKEN
          directory_to_cd_into: "."
          tag: $CIRCLE_TAG
          notes: "This GitHub Release was done automatically by CircleCI"

  hydra_sweeper_test:
    docker:
      - image: 307629990626.dkr.ecr.us-east-1.amazonaws.com/deci/infra/circleci/runner/sg-gpu:<< pipeline.parameters.sg_docker_version >>
    resource_class: deci-ai/sg-gpu-on-premise
    parameters:
      sg_existing_env_path:
        type: string
        default: "/env/persistent_env"
      sg_new_env_name:
        type: string
        default: "${CIRCLE_BUILD_NUM}"
      sg_new_env_python_version:
        type: string
        default: "python3.8"
    steps:
      - checkout
      - run:
          name: install requirements and run recipe tests
          command: |
            << parameters.sg_new_env_python_version >> -m venv << parameters.sg_new_env_name >>
            source << parameters.sg_new_env_name >>/bin/activate
            python3.8 -m pip install --upgrade setuptools pip wheel
            python3.8 -m pip install -r requirements.txt
            python3.8 -m pip install .
            python3.8 -m pip install torch torchvision torchaudio   
            make sweeper_test

      - run:
          name: Remove new environment when failed
          command: "rm -r << parameters.sg_new_env_name >>"
          when: on_fail

      - slack/notify:
          channel: "sg-integration-tests"
          event: fail
          template: basic_fail_1 # see https://github.com/CircleCI-Public/slack-orb/wiki#templates.

  recipe_accuracy_tests:
    docker:
      - image: 307629990626.dkr.ecr.us-east-1.amazonaws.com/deci/infra/circleci/runner/sg-gpu:<< pipeline.parameters.sg_docker_version >>
    resource_class: deci-ai/sg-gpu-on-premise
    parameters:
      sg_existing_env_path:
        type: string
        default: "/env/persistent_env"
      sg_new_env_name:
        type: string
        default: "${CIRCLE_BUILD_NUM}"
      sg_new_env_python_version:
        type: string
        default: "python3.8"
    steps:
      - checkout
      - run:
          name: install requirements and run recipe tests
          command: |
            << parameters.sg_new_env_python_version >> -m venv << parameters.sg_new_env_name >>
            source << parameters.sg_new_env_name >>/bin/activate
            python3.8 -m pip install --upgrade setuptools pip wheel
            python3.8 -m pip install -r requirements.txt
            python3.8 -m pip install .
            python3.8 -m pip install torch torchvision torchaudio
            python3.8 -m pip install pytorch-quantization==2.1.2 --extra-index-url https://pypi.ngc.nvidia.com

            python3.8 tests/verify_min_samples_ddp.py
            python3.8 src/super_gradients/train_from_recipe.py --config-name=coco2017_pose_dekr_w32_no_dc experiment_name=shortened_coco2017_pose_dekr_w32_ap_test batch_size=4 val_batch_size=8 epochs=1 training_hyperparams.lr_warmup_steps=0 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=1000 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cifar10_resnet experiment_name=shortened_cifar10_resnet_accuracy_test epochs=100 training_hyperparams.average_best_models=False multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/examples/convert_recipe_example/convert_recipe_example.py --config-name=cifar10_conversion_params experiment_name=shortened_cifar10_resnet_accuracy_test
            python3.8 src/super_gradients/train_from_recipe.py --config-name=coco2017_yolox experiment_name=shortened_coco2017_yolox_n_map_test architecture=yolox_n training_hyperparams.loss=yolox_fast_loss epochs=10 training_hyperparams.average_best_models=False multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cityscapes_regseg48 experiment_name=shortened_cityscapes_regseg48_iou_test epochs=10 training_hyperparams.average_best_models=False multi_gpu=DDP num_gpus=4
            coverage run --source=super_gradients -m unittest tests/deci_core_recipe_test_suite_runner.py

      - run:
          name: Remove new environment when failed
          command: "rm -r << parameters.sg_new_env_name >>"
          when: on_fail

      - slack/notify:
          channel: "sg-integration-tests"
          event: fail
          template: basic_fail_1 # see https://github.com/CircleCI-Public/slack-orb/wiki#templates.

  recipe_sanity_tests_classification_pt1:
    docker:
      - image: 307629990626.dkr.ecr.us-east-1.amazonaws.com/deci/infra/circleci/runner/sg-gpu:<< pipeline.parameters.sg_docker_version >>
    resource_class: deci-ai/sg-gpu-on-premise
    parameters:
      sg_existing_env_path:
        type: string
        default: "/env/persistent_env"
      sg_new_env_name:
        type: string
        default: "${CIRCLE_BUILD_NUM}"
      sg_new_env_python_version:
        type: string
        default: "python3.8"
    steps:
      - checkout
      - run:
          name: install requirements and run classification sanity tests
          command: |
            << parameters.sg_new_env_python_version >> -m venv << parameters.sg_new_env_name >>
            source << parameters.sg_new_env_name >>/bin/activate
            ls /data/
            python3.8 -m pip install --upgrade setuptools pip wheel
            python3.8 -m pip install -r requirements.txt
            python3.8 -m pip install .
            python3.8 -m pip install torch torchvision torchaudio
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_regnetY architecture=regnetY600 batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_regnetY architecture=regnetY800 batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_repvgg batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_resnet50 batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_vit_base batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
            python3.8 src/super_gradients/train_from_kd_recipe.py --config-name=imagenet_resnet50_kd batch_size=8 val_batch_size=8 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val

      - run:
          name: Remove new environment when failed
          command: "rm -r << parameters.sg_new_env_name >>"
          when: on_fail

      - slack/notify:
          channel: "sg-integration-tests"
          event: fail
          template: basic_fail_1 # see https://github.com/CircleCI-Public/slack-orb/wiki#templates.


  recipe_sanity_tests_classification_pt2:
    docker:
      - image: 307629990626.dkr.ecr.us-east-1.amazonaws.com/deci/infra/circleci/runner/sg-gpu:<< pipeline.parameters.sg_docker_version >>
    resource_class: deci-ai/sg-gpu-on-premise
    parameters:
      sg_existing_env_path:
        type: string
        default: "/env/persistent_env"
      sg_new_env_name:
        type: string
        default: "${CIRCLE_BUILD_NUM}"
      sg_new_env_python_version:
        type: string
        default: "python3.8"
    steps:
      - checkout
      - run:
          name: install requirements and run classification sanity tests
          command: |
            << parameters.sg_new_env_python_version >> -m venv << parameters.sg_new_env_name >>
            source << parameters.sg_new_env_name >>/bin/activate
            python3.8 -m pip install --upgrade setuptools pip wheel
            python3.8 -m pip install -r requirements.txt
            python3.8 -m pip install .
            python3.8 -m pip install torch torchvision torchaudio
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_efficientnet batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_mobilenetv2 batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_mobilenetv3_large batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_mobilenetv3_small batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_regnetY architecture=regnetY200 batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
            python3.8 src/super_gradients/train_from_recipe.py --config-name=imagenet_regnetY architecture=regnetY400 batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4 dataset_params.train_dataset_params.root=/data/Imagenet/train dataset_params.val_dataset_params.root=/data/Imagenet/val
      - run:
          name: Remove new environment when failed
          command: "rm -r << parameters.sg_new_env_name >>"
          when: on_fail
      - slack/notify:
          channel: "sg-integration-tests"
          event: fail
          template: basic_fail_1 # see https://github.com/CircleCI-Public/slack-orb/wiki#templates.

  recipe_sanity_tests_segmentation:
    docker:
      - image: 307629990626.dkr.ecr.us-east-1.amazonaws.com/deci/infra/circleci/runner/sg-gpu:<< pipeline.parameters.sg_docker_version >>
    resource_class: deci-ai/sg-gpu-on-premise
    parameters:
      sg_existing_env_path:
        type: string
        default: "/env/persistent_env"
      sg_new_env_name:
        type: string
        default: "${CIRCLE_BUILD_NUM}"
      sg_new_env_python_version:
        type: string
        default: "python3.8"
    steps:
      - checkout
      - run:
          name: install requirements and run segmentation sanity tests
          command: |
            << parameters.sg_new_env_python_version >> -m venv << parameters.sg_new_env_name >>
            source << parameters.sg_new_env_name >>/bin/activate
            python3.8 -m pip install --upgrade setuptools pip wheel
            python3.8 -m pip install -r requirements.txt
            python3.8 -m pip install .
            python3.8 -m pip install torch torchvision torchaudio
            wget  -O $(pwd)/checkpoints/ddrnet23_slim_bb_imagenet.pth https://deci-pretrained-models.s3.amazonaws.com/ddrnet/imagenet_pt_backbones/ddrnet23_slim_bb_imagenet.pth
            wget  -O $(pwd)/checkpoints/ddrnet23_bb_imagenet.pth https://deci-pretrained-models.s3.amazonaws.com/ddrnet/imagenet_pt_backbones/ddrnet23_bb_imagenet.pth
            wget  -O $(pwd)/checkpoints/ddrnet39_imagenet_pretrained.pth https://deci-pretrained-models.s3.amazonaws.com/ddrnet/imagenet_pt_backbones/ddrnet39_bb_imagenet.pth
            python3.8 src/super_gradients/train_from_recipe.py  --config-name=cityscapes_ddrnet checkpoint_params.checkpoint_path=$(pwd)/checkpoints/ddrnet23_bb_imagenet.pth batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py  --config-name=cityscapes_ddrnet architecture=ddrnet_23_slim checkpoint_params.checkpoint_path=$(pwd)/checkpoints/ddrnet23_slim_bb_imagenet.pth batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cityscapes_ddrnet checkpoint_params.checkpoint_path=$(pwd)/checkpoints/ddrnet39_imagenet_pretrained.pth architecture=ddrnet_39 batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            wget  -O $(pwd)/checkpoints/stdc1_imagenet_pretrained.pth https://deci-pretrained-models.s3.amazonaws.com/stdc_backbones/stdc1_imagenet_pretrained.pth
            wget  -O $(pwd)/checkpoints/stdc2_imagenet_pretrained.pth https://deci-pretrained-models.s3.amazonaws.com/stdc_backbones/stdc2_imagenet_pretrained.pth
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cityscapes_pplite_seg50 checkpoint_params.checkpoint_path=$(pwd)/checkpoints/stdc1_imagenet_pretrained.pth architecture=pp_lite_t_seg batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cityscapes_pplite_seg50 checkpoint_params.checkpoint_path=$(pwd)/checkpoints/stdc2_imagenet_pretrained.pth architecture=pp_lite_b_seg batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cityscapes_pplite_seg75 checkpoint_params.checkpoint_path=$(pwd)/checkpoints/stdc1_imagenet_pretrained.pth architecture=pp_lite_t_seg batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cityscapes_pplite_seg75 checkpoint_params.checkpoint_path=$(pwd)/checkpoints/stdc2_imagenet_pretrained.pth architecture=pp_lite_b_seg batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cityscapes_stdc_seg50 checkpoint_params.checkpoint_path=$(pwd)/checkpoints/stdc1_imagenet_pretrained.pth batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cityscapes_stdc_seg50 checkpoint_params.checkpoint_path=$(pwd)/checkpoints/stdc2_imagenet_pretrained.pth architecture=stdc2_seg batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cityscapes_stdc_seg75 checkpoint_params.checkpoint_path=$(pwd)/checkpoints/stdc1_imagenet_pretrained.pth batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=cityscapes_stdc_seg75 checkpoint_params.checkpoint_path=$(pwd)/checkpoints/stdc2_imagenet_pretrained.pth architecture=stdc2_seg batch_size=3 val_batch_size=3 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
      - run:
          name: Remove new environment when failed
          command: "rm -r << parameters.sg_new_env_name >>"
          when: on_fail

      - slack/notify:
          channel: "sg-integration-tests"
          event: fail
          template: basic_fail_1 # see https://github.com/CircleCI-Public/slack-orb/wiki#templates.

  recipe_sanity_tests_detection:
    docker:
      - image: 307629990626.dkr.ecr.us-east-1.amazonaws.com/deci/infra/circleci/runner/sg-gpu:<< pipeline.parameters.sg_docker_version >>
    resource_class: deci-ai/sg-gpu-on-premise
    parameters:
      sg_existing_env_path:
        type: string
        default: "/env/persistent_env"
      sg_new_env_name:
        type: string
        default: "${CIRCLE_BUILD_NUM}"
      sg_new_env_python_version:
        type: string
        default: "python3.8"
    steps:
      - checkout
      - run:
          name: install requirements and run detection sanity tests
          command: |
            << parameters.sg_new_env_python_version >> -m venv << parameters.sg_new_env_name >>
            source << parameters.sg_new_env_name >>/bin/activate
            python3.8 -m pip install --upgrade setuptools pip wheel
            python3.8 -m pip install -r requirements.txt
            python3.8 -m pip install .
            python3.8 -m pip install torch torchvision torchaudio
            python3.8 src/super_gradients/train_from_recipe.py --config-name=coco2017_ssd_lite_mobilenet_v2 batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_n batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_t batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_s batch_size=8 val_batch_size=16 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_m batch_size=8 val_batch_size=8 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4
            python3.8 src/super_gradients/train_from_recipe.py --config-name=coco2017_yolox architecture=yolox_l batch_size=4 val_batch_size=8 epochs=1 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=100 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4

      - run:
          name: Remove new environment when failed
          command: "rm -r << parameters.sg_new_env_name >>"
          when: on_fail
      - slack/notify:
          channel: "sg-integration-tests"
          event: fail
          template: basic_fail_1 # see https://github.com/CircleCI-Public/slack-orb/wiki#templates.

  recipe_sanity_tests_pose_estimation:
    docker:
      - image: 307629990626.dkr.ecr.us-east-1.amazonaws.com/deci/infra/circleci/runner/sg-gpu:<< pipeline.parameters.sg_docker_version >>
    resource_class: deci-ai/sg-gpu-on-premise
    parameters:
      sg_existing_env_path:
        type: string
        default: "/env/persistent_env"
      sg_new_env_name:
        type: string
        default: "${CIRCLE_BUILD_NUM}"
      sg_new_env_python_version:
        type: string
        default: "python3.8"
    steps:
      - checkout
      - run:
          name: install requirements and run pose estimation sanity tests
          command: |
            << parameters.sg_new_env_python_version >> -m venv << parameters.sg_new_env_name >>
            source << parameters.sg_new_env_name >>/bin/activate
            python3.8 -m pip install --upgrade setuptools pip wheel
            python3.8 -m pip install -r requirements.txt
            python3.8 -m pip install .
            python3.8 -m pip install torch torchvision torchaudio
            python3.8 src/super_gradients/train_from_recipe.py --config-name=coco2017_pose_dekr_w32_no_dc experiment_name=shortened_coco2017_pose_dekr_w32_ap_test batch_size=4 val_batch_size=8 epochs=1 training_hyperparams.lr_warmup_steps=0 training_hyperparams.average_best_models=False training_hyperparams.max_train_batches=1000 training_hyperparams.max_valid_batches=100 multi_gpu=DDP num_gpus=4

      - run:
          name: Remove new environment when failed
          command: "rm -r << parameters.sg_new_env_name >>"
          when: on_fail
      - slack/notify:
          channel: "sg-integration-tests"
          event: fail
          template: basic_fail_1 # see https://github.com/CircleCI-Public/slack-orb/wiki#templates.

  docker-build-and-publish-branch:
    docker:
      - image: cimg/base:stable-20.04
    parameters:
      repo_name:
        type: string
        default: "deci/super-gradients"
      docker_context:
        type: string
        default: "."
      additional_tags:
        type: string
        default: ''
    steps:
      - setup_remote_docker:
          version: 20.10.7
          docker_layer_caching: true
      - deci-common/container_image_lint_tag:
          image_tag: "${CIRCLE_BRANCH}"
      - run:
          command: |
            ADDITIONAL_TAGS="<< parameters.additional_tags >>"
            echo "export ADDITIONAL_TAGS=${ADDITIONAL_TAGS}" >> $BASH_ENV
      - run:
          command: |
            source $BASH_ENV
            echo "$CONTAINER_LINT_TAG"
            echo "$ADDITIONAL_TAGS"
      - when:
          condition: << pipeline.parameters.ad_hoc_container_build_code_only >>
          steps:
            - build_and_publish_command:
                repo_name: << parameters.repo_name >>
                docker_context: << parameters.docker_context >>
                image_tag: $CONTAINER_LINT_TAG
                additional_tags: $ADDITIONAL_TAGS
                dockerfile: 'scripts/Dockerfile.branch.code'
                build_args: "BASE_TAG=$CONTAINER_LINT_TAG"
      - unless:
          condition: << pipeline.parameters.ad_hoc_container_build_code_only >>
          steps:
            - build_and_publish_command:
                repo_name: << parameters.repo_name >>
                docker_context: << parameters.docker_context >>
                image_tag: $CONTAINER_LINT_TAG
                additional_tags: $ADDITIONAL_TAGS
                dockerfile: 'scripts/Dockerfile.branch'

workflows:
  sanity_tests:
    when:  << pipeline.parameters.run_sanity_tests_flow >>
    jobs:
      - hydra_sweeper_test:
          context:
            - slack
      - recipe_sanity_tests_classification_pt1:
          context:
            - slack
      - recipe_sanity_tests_classification_pt2:
          context:
            - slack
      - recipe_sanity_tests_segmentation:
          context:
            - slack
      - recipe_sanity_tests_detection:
          context:
            - slack
      - recipe_sanity_tests_pose_estimation:
          context:
            - slack
      - recipe_accuracy_tests:
          context:
            - slack

  release:
    unless:
      or:
        - << pipeline.parameters.ad_hoc_container_build >>
        - << pipeline.parameters.ad_hoc_container_build_code_only >>
    jobs:
      - deci-common/persist_version_info:
          version_override: $CIRCLE_TAG
          <<: *release_tag_filter
      - deci-common/codeartifact_login:
          name: "login_to_codeartifact_release"
          repo_name: "deci-packages"
          <<: *release_tag_filter
      - build:
          name: "build3.8"
          py_version: "3.8"
          package_name: "super-gradients"
          requires:
            - deci-common/persist_version_info
            - login_to_codeartifact_release
          <<: *release_tag_filter
      - hydra_sweeper_test:
          context:
            - slack
          <<: *release_tag_filter
      - recipe_sanity_tests_classification_pt1:
          context:
            - slack
          <<: *release_tag_filter
      - recipe_sanity_tests_classification_pt2:
          context:
            - slack
          <<: *release_tag_filter
      - recipe_sanity_tests_segmentation:
          context:
            - slack
          <<: *release_tag_filter
      - recipe_sanity_tests_detection:
          context:
            - slack
          <<: *release_tag_filter
      - recipe_sanity_tests_pose_estimation:
          context:
            - slack
          <<: *release_tag_filter
      - recipe_accuracy_tests:
          context:
            - slack
          <<: *release_tag_filter
      - release_version:
          py_version: "3.8"
          requires:
            - "build3.8"
            - hydra_sweeper_test
            - recipe_accuracy_tests
            - recipe_sanity_tests_classification_pt1
            - recipe_sanity_tests_classification_pt2
            - recipe_sanity_tests_segmentation
            - recipe_sanity_tests_detection
            - recipe_sanity_tests_pose_estimation
          <<: *release_tag_filter
      - deci-common/pip_upload_package_from_codeartifact_to_global_pypi:
          package_name: "super-gradients"
          name: "upload_super_gradients_to_pypi"
          version: $CIRCLE_TAG
          requires:
            - "release_version"
          context:
            - pypi-supergradients
          <<: *release_tag_filter


  build_and_deploy:
    unless:
      or:
        - << pipeline.parameters.ad_hoc_container_build >>
        - << pipeline.parameters.ad_hoc_container_build_code_only >>
    jobs:
      - deci-common/persist_version_info:
          use_rc: true
          use_beta: false
          version_override: ""

      - build:
          name: "build<< pipeline.parameters.build_py_version >>"
          py_version:  << pipeline.parameters.build_py_version >>
          package_name: "super-gradients"
          requires:
            - deci-common/persist_version_info

      - breaking_change_check:
          name: "breaking-change-check"
          py_version: << pipeline.parameters.build_py_version >>
          requires:
            - deci-common/persist_version_info

      - deci-common/codeartifact_login:
          repo_name: "deci-packages"
          <<: *release_candidate_filter

      - release_candidate: # happens on merge
          py_version: "<< pipeline.parameters.build_py_version >>"
          requires:
            - "build<< pipeline.parameters.build_py_version >>"
            - deci-common/codeartifact_login
          <<: *release_candidate_filter

  SG_docker:
    unless:
      or:
        - << pipeline.parameters.ad_hoc_container_build >>
        - << pipeline.parameters.ad_hoc_container_build_code_only >>
    jobs:
      - change_rc_to_b: # works on release candidate creation
          <<: *release_candidate_tag_filter
      - build_and_publish_sg_container:  # works on release candidate creation
          requires:
            - "change_rc_to_b"
          <<: *release_candidate_tag_filter
      - testing_supergradients_docker_image:  # works on release candidate creation
          image_repo: '307629990626.dkr.ecr.us-east-1.amazonaws.com/deci/super-gradients'
          requires:
            - "build_and_publish_sg_container"
            - "change_rc_to_b"
          <<: *release_candidate_tag_filter
      - add_rc_tag_to_beta: # works on release candidate creation for ECR Repo
          requires:
            - "testing_supergradients_docker_image"
            - "change_rc_to_b"
          <<: *release_candidate_tag_filter
      - find_rc_tag_per_sha: # works on release
          <<: *release_tag_filter
      - add_release_tag_to_rc: # works on release
          requires:
            - "find_rc_tag_per_sha"
          <<: *release_tag_filter
      - slack/on-hold:
          context: slack
          channel: "sg-integration-tests"
          requires:
           - "add_release_tag_to_rc"
          <<: *release_tag_filter
      - hold-sg-public-release:  # works on release
          type: approval
          requires:
           - "slack/on-hold"
          <<: *release_tag_filter
      - docker/publish:  # works on release
          executor:
              image: cimg/base
              tag: stable-20.04
              name: docker/docker
          image: deciai/super-gradients
          remote-docker-version: << pipeline.parameters.remote_docker_version >>
          update-description: true
          use-buildkit: true
          use-remote-docker: true
          use-docker-credentials-store: true
          path: ./scripts
          readme: ../README.md
          tag: latest,${CIRCLE_TAG}
          extra_build_args: '--build-arg VERSION=${CIRCLE_TAG}' #building from scratch as it faster and cheaper than download and retag
          requires:
            - "hold-sg-public-release"
          <<: *release_tag_filter
      - docker/publish: # works on release
          executor:
              image: cimg/base
              tag: stable-20.04
              name: docker/docker
          image: deciai/super-gradients
          remote-docker-version: << pipeline.parameters.remote_docker_version >>
          update-description: false
          use-buildkit: true
          use-remote-docker: true
          use-docker-credentials-store: true
          path: ./scripts
          tag: ${CIRCLE_TAG}-runtime
          extra_build_args: '--build-arg VERSION=${CIRCLE_TAG} --build-arg DOCKER_IMAGE_TAG=11.3.1-runtime-ubuntu20.04'
          requires:
            - "hold-sg-public-release"
          <<: *release_tag_filter
  build-and-push-container-flow:
    when: << pipeline.parameters.ad_hoc_container_build >>
    jobs:
      - docker-build-and-publish-branch
  build-and-push-container-code-only-flow:
    when: << pipeline.parameters.ad_hoc_container_build_code_only >>
    jobs:
      - docker-build-and-publish-branch