Skip to content

Latest commit

 

History

History
71 lines (60 loc) · 1.84 KB

README.md

File metadata and controls

71 lines (60 loc) · 1.84 KB

Coiled Run GitHub Action

This GitHub Action runs a command on a Coiled cluster by adding the following steps to your workflow. For more details see our CLI job docs.

  - name: Coiled Login
    uses: coiled/login-action@v1
    with:
      token: ${{ secrets.COILED_API_TOKEN }}
  - name: Coiled Run
    uses: coiled/run-action@v1
    with:
      command: "echo 'Hello, Coiled!'"

Examples

Run a command on a Coiled cluster

  - name: Coiled Login
    uses: coiled/login-action@v1
    with:
      token: ${{ secrets.COILED_API_TOKEN }}
  - name: Coiled Run
    uses: coiled/run-action@v1
    with:
      command: "echo 'Hello, Coiled!'"

Use a GPU VM to train a model with Python using package sync

  - name: Coiled Login
    uses: coiled/login-action@v1
    with:
      token: ${{ secrets.COILED_API_TOKEN }}
  - name: Coiled Run
    uses: coiled/run-action@v1
    with:
      command: "python train.py"
      vm-type: "g4dn.xlarge"

Run a Python script that uses a manual software environment

  - name: Coiled Login
    uses: coiled/login-action@v1
    with:
      token: ${{ secrets.COILED_API_TOKEN }}
  - name: Coiled Run
    uses: coiled/run-action@v1
    with:
      command: "python my_script.py"
      software: "my-env"

Run a command in a Docker container

  - name: Coiled Login
    uses: coiled/login-action@v1
    with:
      token: ${{ secrets.COILED_API_TOKEN }}
  - name: Coiled Run
    uses: coiled/run-action@v1
    with:
      command: "python rapids_script.py"
      container: "nvcr.io/nvidia/rapidsai/base:23.08-cuda11.8-py3.10"