Manual Docker Build for Custom Environments on Gradescope

1.Clone the Repository:

```
git clone https://github.com/jlm429/gradescope_autograder
```

2.Customize the Docker Environment:

• Select a base image: See Gradescope Docker Hub repository.
• Modify Dockerfile or setup.sh: Customize the environment by editing the Dockerfile or the setup.sh script.
  • Example:

    • In the Dockerfile: Add commands to update the package list and install necessary packages.

      ```
      RUN apt-get update && \
          apt-get install -yq --no-install-recommends \
          python3-pip \
          && apt-get clean
      ```
      

    • In setup.sh: Add commands to upgrade pip and install required Python packages.

      ```
      #!/usr/bin/env bash
      pip install --no-cache-dir --upgrade pip 
      pip install --no-cache-dir torch
      pip install -r /autograder/source/requirements.txt
      ```
      

3.Change to the Docker Directory and Build the Image:

```
cd gradescope_autograder/docker
docker build -t username/image_name:tag .
```

4.Push the Image to DockerHub:

```
docker push username/image_name:tag 
```

5.Use the Image on Gradescope:

• Pull the Docker image on Gradescope by entering the DockerHub username/image_name:tag.
• Submit example_submit.py for testing.

6.Optional: Test the Build Locally:

```
docker run --rm -v /path/to/submission:/autograder/submission -v /path/to/results:/autograder/results username/image_name:tag /autograder/run_autograder && cat /path/to/results/results.json
```

For more detailed instructions, refer to the Gradescope Autograders Documentation.