GitHub Action For Running Experiments With ClearML

Train models easily with ClearML, directly from your repo!

This action helps to run experiments with , directly from Github. Just comment from any issue or pull request with

• /train-model branch <brach name>
• /train-model tag <tag_name>
• /train-model commit <commit_id>

Usage

Workflow Example

This will add an action to your workflow that will clone a ClearML Task TASK_ID and will enqueue it to selected Queue (QUEUE_NAME input parameter).

Works both in github issues and github pull requests comments.

name: Train model
on: [issue_comment]

jobs:
  train-model:
      if: contains(github.event.comment.body, '/train-model')
      runs-on: ubuntu-latest
      steps:
        - name: Train model
          uses: allegroai/clearml-train-model@master
          id: train
          with:
            CLEARML_API_ACCESS_KEY: ${{ secrets.ACCESS_KEY }}
            CLEARML_API_SECRET_KEY: ${{ secrets.SECRET_KEY }}
            CLEARML_API_HOST: ${{ secrets.CLEARML_API_HOST }}
            TASK_ID: "e4623efdfa1d461e9101615728fdc52e"
            QUEUE_NAME: "train_queue"
          env:
            GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        # Example how you can use outputs from the train action
        - name: Print task status
          run: |
            echo "Task stats is ${TASK_STATUS} for task ${CLONED_TASK}"
          env:
            TASK_STATUS: ${{ steps.train.outputs.TASK_STATUS }}
            CLONED_TASK: ${{ steps.train.outputs.CLONED_TASK }}

Inputs

Mandatory

1. CLEARML_API_ACCESS_KEY: Your ClearML api access key. You can find it in your clearml.conf file under api.credentials.access_key section, read more.
2. CLEARML_API_SECRET_KEY: Your ClearML api secret key. You can find it in your clearml.conf file under api.credentials.secret_key section, read more.
3. CLEARML_API_HOST: The ClearML api server address. You can find it in your clearml.conf file under api.api_server section, read more.
4. TASK_ID: Id of the task you would like to clone.

Optional

1. QUEUE_NAME: Queue for the cloned task (default value: default). You can read more about queues here.

Outputs

1. CLONED_TASK: The cloned task id.
2. TASK_STATUS: The cloned task status (not updating).