Docker for ML reproducibility in MLOps - Practice Problems & Coding Challenges

Challenge - 5 Problems

💻 Command Output

intermediate

What is the output of this Docker command?

You run the command docker run --rm python:3.9-slim python -c "print(2+3)". What will be the output?

MLOps

docker run --rm python:3.9-slim python -c "print(2+3)"

A 5

B python:3.9-slim: command not found

C Error: No such image: python:3.9-slim

D 2+3

💡 Hint

Think about what the python command inside the container does.

The command runs a Python one-liner inside the official python:3.9-slim container, printing the sum 5.

🧠 Conceptual

intermediate

Why use Docker for ML reproducibility?

Which of the following best explains why Docker helps with reproducibility in machine learning projects?

A Docker provides a cloud service to train ML models faster.

B Docker automatically improves ML model accuracy by optimizing code.

C Docker replaces the need for data preprocessing in ML pipelines.

D Docker packages code, dependencies, and environment so ML runs the same everywhere.

💡 Hint

Think about what Docker containers include.

Docker containers bundle the code, libraries, and environment, ensuring consistent runs across different machines.

❓ Configuration

advanced

Which Dockerfile snippet correctly installs Python dependencies for ML?

You want to create a Docker image for your ML project. Which Dockerfile snippet correctly installs dependencies from requirements.txt?

MLOps

FROM python:3.8-slim
COPY requirements.txt ./
RUN ???
COPY . ./app
WORKDIR /app
CMD ["python", "train.py"]

A RUN pip install -r ./app/requirements.txt

B RUN pip install requirements.txt

C RUN pip install -r requirements.txt

D RUN install pip -r requirements.txt

💡 Hint

The command must tell pip to read the requirements file.

The correct syntax to install dependencies from a requirements file is 'pip install -r requirements.txt'.

❓ Troubleshoot

advanced

Why does this Docker container fail to find the ML script?

You build and run a Docker container but get an error: python: can't open file 'train.py': [Errno 2] No such file or directory. The Dockerfile is:

FROM python:3.8-slim
COPY . /app
WORKDIR /app
CMD ["python", "train.py"]

What is the most likely cause?

A The WORKDIR is set after CMD, so the command runs in the wrong directory.

B The <code>train.py</code> file is missing in the build context directory.

C The COPY command should be <code>COPY /app .</code> instead.

D Python is not installed in the python:3.8-slim image.

💡 Hint

Check what files are available when building the image.

If train.py is not in the folder used to build the image, it won't be copied into /app, causing the error.

🔀 Workflow

expert

Order the steps to create a Docker image for ML reproducibility

Arrange these steps in the correct order to create and run a Docker container for an ML project:

A 4,1,2,3

B 1,4,2,3

C 4,2,1,3

D 1,2,4,3

💡 Hint

Think about preparing code before writing Dockerfile and building image.

First prepare code, then write Dockerfile, build image, and finally run container.