Logging artifacts and models in MLOps - Practice Problems & Coding Challenges

Challenge - 5 Problems

💻 Command Output

intermediate

What is the output of this MLflow artifact logging command?

Consider the following Python code snippet using MLflow to log an artifact file. What will be the output printed after running this code?

MLOps

import mlflow
with open('model_info.txt', 'w') as f:
    f.write('Model version 1.0')
with mlflow.start_run() as run:
    mlflow.log_artifact('model_info.txt')
    print(run.info.artifact_uri)

A file:///path/to/mlruns/0/<run_id>/artifacts

B s3://mlflow-bucket/artifacts/<run_id>

C http://localhost:5000/api/2.0/mlflow/artifacts/<run_id>

D Error: artifact_uri attribute not found

💡 Hint

The artifact_uri shows the local path where artifacts are stored during the run.

MLflow stores artifacts locally by default under the mlruns directory. The artifact_uri attribute gives the local file path for the current run's artifacts.

🧠 Conceptual

intermediate

Which MLflow command logs a model for later deployment?

You want to save a trained scikit-learn model in MLflow so it can be deployed later. Which command correctly logs the model?

A mlflow.log_metric('model', model)

B mlflow.sklearn.log_model(model, 'model')

C mlflow.log_param('model', model)

D mlflow.log_artifact(model, 'model')

💡 Hint

Logging a model uses a specific MLflow flavor function.

mlflow.sklearn.log_model saves the model in a format MLflow can use for deployment. log_artifact is for files, log_param and log_metric are for parameters and metrics.

❓ Troubleshoot

advanced

Why does MLflow fail to log artifacts in this code?

You run this code but no artifacts appear in the MLflow UI. What is the most likely cause?

MLOps

import mlflow
with mlflow.start_run():
    mlflow.log_artifact('output.txt')
# output.txt is created after the run block

A The artifact file does not exist when log_artifact is called

B output.txt must be a directory, not a file

C log_artifact only works outside the run context

D MLflow requires explicit run ID to log artifacts

💡 Hint

Check when the file is created relative to when it is logged.

MLflow cannot log a file that does not exist at the time log_artifact is called. The file must be created before logging.

🔀 Workflow

advanced

Order the steps to log a model and its metrics in MLflow

Arrange these steps in the correct order to log a model and its evaluation metrics using MLflow.

A 1,3,2,4,5

B 1,2,3,4,5

C 2,1,4,3,5

D 2,1,3,4,5

💡 Hint

You must start the run before logging anything.

The correct workflow is to start the run, train the model, log the model, log metrics, then end the run.

✅ Best Practice

expert

What is the best practice for versioning models in MLflow?

You want to keep track of multiple versions of your models in MLflow for easy rollback and comparison. Which approach is best?

A Store models in separate folders outside MLflow

B Manually rename model files with version numbers before logging

C Use MLflow Model Registry to register and version models

D Log models without versioning and rely on timestamps

💡 Hint

MLflow provides a built-in system for model versioning.

MLflow Model Registry is designed to manage model versions, stages, and metadata centrally. Manual renaming or external storage is error-prone.