Determine the Differences Between Jupyter Notebooks and Pipelines CERTIFIED-AI-EXPERT Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A company needs to retrain a machine learning model every week using newly collected data. The training process involves data validation, feature engineering, model training, and model evaluation. The process must run automatically without human intervention and produce a report of the model's performance. Which approach is most appropriate?

   • AA Jupyter notebook that is manually run by an engineer every week.
   • BA single Python script that performs all steps sequentially, run manually.
   • CA pipeline that defines each step as a separate component with explicit inputs and outputs, scheduled to run weekly.
   • DA Jupyter notebook that is scheduled to run automatically via a cron job.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   When is it most appropriate to use a Jupyter notebook?

   • AWhen you need to ensure reproducible runs with versioned data and code.
   • BWhen you need to run a scheduled, unattended data processing job.
   • CWhen you need to deploy a model to a production REST API.
   • DWhen you are exploring data and iterating on analysis code interactively.
   Reveal Solution Discussion Select an answer first
4. 8expert · hard

   A team is developing a model that requires a long-running training job (several hours) on a GPU cluster. The team has been prototyping in a Jupyter notebook. They need to run the training job on a dedicated GPU node and ensure that the job can be resumed from a checkpoint if the node fails. The team also wants to track the exact code version and hyperparameters used for each run. What is the most appropriate approach?

   • AWrite a single monolithic Python script that runs the training and saves checkpoints, and run it manually on the GPU node.
   • BConvert the training logic into a pipeline step that reads hyperparameters from a config file, saves checkpoints, and logs the code version and parameters.
   • CRun the training directly in the notebook on the GPU node and save checkpoints manually.
   • DUse a notebook with papermill to parameterize and execute the training on the GPU node.
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A team of researchers is collaborating on a shared Jupyter notebook to analyze experimental results. They are using Git for version control. The team notices that the notebook file often shows large, confusing diffs even when only a small code change is made. What is the most likely cause of this issue?

   • AThe notebook file stores code, outputs, and metadata in a single JSON structure, so any change to outputs or metadata creates a large diff.
   • BThe notebook is too large to be tracked by Git, so Git shows a large diff for any change.
   • CThe team is using a Git client that is not compatible with Jupyter notebooks.
   • DGit does not support JSON files, so it treats the notebook as a binary file and shows a large diff.
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   How does a notebook maintain state across cells?

   • ABy saving all variables to disk after each cell execution.
   • BBy re-executing all previous cells before each new cell.
   • CBy keeping variables in the kernel's memory between cell executions.
   • DBy storing state in a global configuration file.
   Reveal Solution Discussion Select an answer first